* License.....: MIT
*/
-static int hash_comp (const u32 d1[4], __global u32 *d2)
+/**
+ * pure scalar functions
+ */
+
+inline int hash_comp (const u32 d1[4], __global u32 *d2)
{
if (d1[3] > d2[DGST_R3]) return ( 1);
if (d1[3] < d2[DGST_R3]) return (-1);
return (0);
}
-static int find_hash (const u32 digest[4], const u32 digests_cnt, __global digest_t *digests_buf)
+inline int find_hash (const u32 digest[4], const u32 digests_cnt, __global digest_t *digests_buf)
{
for (u32 l = 0, r = digests_cnt; r; r >>= 1)
{
return (-1);
}
-static u32 check_bitmap (__global u32 *bitmap, const u32 bitmap_mask, const u32 bitmap_shift, const u32 digest)
+inline u32 check_bitmap (__global u32 *bitmap, const u32 bitmap_mask, const u32 bitmap_shift, const u32 digest)
{
return (bitmap[(digest >> bitmap_shift) & bitmap_mask] & (1 << (digest & 0x1f)));
}
-static u32 check (const u32 digest[2], __global u32 *bitmap_s1_a, __global u32 *bitmap_s1_b, __global u32 *bitmap_s1_c, __global u32 *bitmap_s1_d, __global u32 *bitmap_s2_a, __global u32 *bitmap_s2_b, __global u32 *bitmap_s2_c, __global u32 *bitmap_s2_d, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2)
+inline u32 check (const u32 digest[2], __global u32 *bitmap_s1_a, __global u32 *bitmap_s1_b, __global u32 *bitmap_s1_c, __global u32 *bitmap_s1_d, __global u32 *bitmap_s2_a, __global u32 *bitmap_s2_b, __global u32 *bitmap_s2_c, __global u32 *bitmap_s2_d, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2)
{
if (check_bitmap (bitmap_s1_a, bitmap_mask, bitmap_shift1, digest[0]) == 0) return (0);
if (check_bitmap (bitmap_s1_b, bitmap_mask, bitmap_shift1, digest[1]) == 0) return (0);
return (1);
}
-static void mark_hash (__global plain_t *plains_buf, __global u32 *hashes_shown, const int hash_pos, const u32 gid, const u32 il_pos)
+inline void mark_hash (__global plain_t *plains_buf, __global u32 *hashes_shown, const int hash_pos, const u32 gid, const u32 il_pos)
{
hashes_shown[hash_pos] = 1;
plains_buf[hash_pos].il_pos = il_pos;
}
-static void truncate_block (u32 w[4], const u32 len)
+/**
+ * vector functions
+ */
+
+inline void truncate_block (u32x w[4], const u32 len)
{
switch (len)
{
}
}
-static void make_unicode_S (const u32 in[4], u32 out1[4], u32 out2[4])
-{
- #ifdef IS_NV
- out2[3] = __byte_perm_S (in[3], 0, 0x7372);
- out2[2] = __byte_perm_S (in[3], 0, 0x7170);
- out2[1] = __byte_perm_S (in[2], 0, 0x7372);
- out2[0] = __byte_perm_S (in[2], 0, 0x7170);
- out1[3] = __byte_perm_S (in[1], 0, 0x7372);
- out1[2] = __byte_perm_S (in[1], 0, 0x7170);
- out1[1] = __byte_perm_S (in[0], 0, 0x7372);
- out1[0] = __byte_perm_S (in[0], 0, 0x7170);
- #endif
-
- #if defined IS_AMD || defined IS_GENERIC
- out2[3] = ((in[3] >> 8) & 0x00FF0000) | ((in[3] >> 16) & 0x000000FF);
- out2[2] = ((in[3] << 8) & 0x00FF0000) | ((in[3] >> 0) & 0x000000FF);
- out2[1] = ((in[2] >> 8) & 0x00FF0000) | ((in[2] >> 16) & 0x000000FF);
- out2[0] = ((in[2] << 8) & 0x00FF0000) | ((in[2] >> 0) & 0x000000FF);
- out1[3] = ((in[1] >> 8) & 0x00FF0000) | ((in[1] >> 16) & 0x000000FF);
- out1[2] = ((in[1] << 8) & 0x00FF0000) | ((in[1] >> 0) & 0x000000FF);
- out1[1] = ((in[0] >> 8) & 0x00FF0000) | ((in[0] >> 16) & 0x000000FF);
- out1[0] = ((in[0] << 8) & 0x00FF0000) | ((in[0] >> 0) & 0x000000FF);
- #endif
-}
-
-static void make_unicode (const u32x in[4], u32x out1[4], u32x out2[4])
+inline void make_unicode (const u32x in[4], u32x out1[4], u32x out2[4])
{
#ifdef IS_NV
out2[3] = __byte_perm (in[3], 0, 0x7372);
#endif
}
-static void undo_unicode_S (const u32 in1[4], const u32 in2[4], u32 out[4])
-{
- #ifdef IS_NV
- out[0] = __byte_perm_S (in1[0], in1[1], 0x6420);
- out[1] = __byte_perm_S (in1[2], in1[3], 0x6420);
- out[2] = __byte_perm_S (in2[0], in2[1], 0x6420);
- out[3] = __byte_perm_S (in2[2], in2[3], 0x6420);
- #endif
-
- #if defined IS_AMD || defined IS_GENERIC
- out[0] = ((in1[0] & 0x000000ff) >> 0) | ((in1[0] & 0x00ff0000) >> 8)
- | ((in1[1] & 0x000000ff) << 16) | ((in1[1] & 0x00ff0000) << 8);
- out[1] = ((in1[2] & 0x000000ff) >> 0) | ((in1[2] & 0x00ff0000) >> 8)
- | ((in1[3] & 0x000000ff) << 16) | ((in1[3] & 0x00ff0000) << 8);
- out[2] = ((in2[0] & 0x000000ff) >> 0) | ((in2[0] & 0x00ff0000) >> 8)
- | ((in2[1] & 0x000000ff) << 16) | ((in2[1] & 0x00ff0000) << 8);
- out[3] = ((in2[2] & 0x000000ff) >> 0) | ((in2[2] & 0x00ff0000) >> 8)
- | ((in2[3] & 0x000000ff) << 16) | ((in2[3] & 0x00ff0000) << 8);
- #endif
-}
-
-static void undo_unicode (const u32x in1[4], const u32x in2[4], u32x out[4])
+inline void undo_unicode (const u32x in1[4], const u32x in2[4], u32x out[4])
{
#ifdef IS_NV
out[0] = __byte_perm (in1[0], in1[1], 0x6420);
#endif
}
-static void append_0x01_1x4 (u32 w0[4], const u32 offset)
+inline void append_0x01_1x4 (u32x w0[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x01_2x4 (u32 w0[4], u32 w1[4], const u32 offset)
+inline void append_0x01_2x4 (u32x w0[4], u32x w1[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x01_3x4 (u32 w0[4], u32 w1[4], u32 w2[4], const u32 offset)
+inline void append_0x01_3x4 (u32x w0[4], u32x w1[4], u32x w2[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x01_4x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+inline void append_0x01_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x01_8x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const u32 offset)
+inline void append_0x01_8x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x02_1x4 (u32 w0[4], const u32 offset)
+inline void append_0x02_1x4 (u32x w0[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x02_2x4 (u32 w0[4], u32 w1[4], const u32 offset)
+inline void append_0x02_2x4 (u32x w0[4], u32x w1[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x02_3x4 (u32 w0[4], u32 w1[4], u32 w2[4], const u32 offset)
+inline void append_0x02_3x4 (u32x w0[4], u32x w1[4], u32x w2[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x02_4x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+inline void append_0x02_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x02_8x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const u32 offset)
+inline void append_0x02_8x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_1x4 (u32 w0[4], const u32 offset)
+inline void append_0x80_1x4 (u32x w0[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_2x4 (u32x w0[4], u32x w1[4], const u32 offset)
+inline void append_0x80_2x4 (u32x w0[4], u32x w1[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_3x4 (u32 w0[4], u32 w1[4], u32 w2[4], const u32 offset)
+inline void append_0x80_3x4 (u32x w0[4], u32x w1[4], u32x w2[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_4x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+inline void append_0x80_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_8x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 w4[4], u32 w5[4], u32 w6[4], u32 w7[4], const u32 offset)
+inline void append_0x80_8x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], const u32 offset)
{
switch (offset)
{
}
}
-static void append_0x80_1x16 (u32 w[16], const u32 offset)
+inline void append_0x80_1x16 (u32x w[16], const u32 offset)
{
switch (offset)
{
}
}
-static void switch_buffer_by_offset_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+inline void switch_buffer_by_offset_le (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
{
#if defined IS_AMD || defined IS_GENERIC
const int offset_mod_4 = offset & 3;
switch (offset / 4)
{
case 0:
- w3[2] = amd_bytealign_S ( 0, w3[1], offset_minus_4);
- w3[1] = amd_bytealign_S (w3[1], w3[0], offset_minus_4);
- w3[0] = amd_bytealign_S (w3[0], w2[3], offset_minus_4);
- w2[3] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
- w2[2] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
- w2[1] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
- w2[0] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w1[3] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w1[2] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w1[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w1[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w0[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w0[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w0[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w0[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w3[1], offset_minus_4);
+ w3[1] = amd_bytealign (w3[1], w3[0], offset_minus_4);
+ w3[0] = amd_bytealign (w3[0], w2[3], offset_minus_4);
+ w2[3] = amd_bytealign (w2[3], w2[2], offset_minus_4);
+ w2[2] = amd_bytealign (w2[2], w2[1], offset_minus_4);
+ w2[1] = amd_bytealign (w2[1], w2[0], offset_minus_4);
+ w2[0] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w1[3] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w1[2] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w1[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w1[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w0[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w0[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w0[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w0[0] = amd_bytealign (w0[0], 0, offset_minus_4);
if (offset_mod_4 == 0)
{
break;
case 1:
- w3[2] = amd_bytealign_S ( 0, w3[0], offset_minus_4);
- w3[1] = amd_bytealign_S (w3[0], w2[3], offset_minus_4);
- w3[0] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
- w2[3] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
- w2[2] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
- w2[1] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w2[0] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w1[3] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w1[2] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w1[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w1[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w0[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w0[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w0[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w3[0], offset_minus_4);
+ w3[1] = amd_bytealign (w3[0], w2[3], offset_minus_4);
+ w3[0] = amd_bytealign (w2[3], w2[2], offset_minus_4);
+ w2[3] = amd_bytealign (w2[2], w2[1], offset_minus_4);
+ w2[2] = amd_bytealign (w2[1], w2[0], offset_minus_4);
+ w2[1] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w2[0] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w1[3] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w1[2] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w1[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w1[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w0[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w0[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w0[1] = amd_bytealign (w0[0], 0, offset_minus_4);
w0[0] = 0;
if (offset_mod_4 == 0)
break;
case 2:
- w3[2] = amd_bytealign_S ( 0, w2[3], offset_minus_4);
- w3[1] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
- w3[0] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
- w2[3] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
- w2[2] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w2[1] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w2[0] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w1[3] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w1[2] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w1[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w1[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w0[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w0[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w2[3], offset_minus_4);
+ w3[1] = amd_bytealign (w2[3], w2[2], offset_minus_4);
+ w3[0] = amd_bytealign (w2[2], w2[1], offset_minus_4);
+ w2[3] = amd_bytealign (w2[1], w2[0], offset_minus_4);
+ w2[2] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w2[1] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w2[0] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w1[3] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w1[2] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w1[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w1[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w0[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w0[2] = amd_bytealign (w0[0], 0, offset_minus_4);
w0[1] = 0;
w0[0] = 0;
break;
case 3:
- w3[2] = amd_bytealign_S ( 0, w2[2], offset_minus_4);
- w3[1] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
- w3[0] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
- w2[3] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w2[2] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w2[1] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w2[0] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w1[3] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w1[2] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w1[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w1[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w0[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w2[2], offset_minus_4);
+ w3[1] = amd_bytealign (w2[2], w2[1], offset_minus_4);
+ w3[0] = amd_bytealign (w2[1], w2[0], offset_minus_4);
+ w2[3] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w2[2] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w2[1] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w2[0] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w1[3] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w1[2] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w1[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w1[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w0[3] = amd_bytealign (w0[0], 0, offset_minus_4);
w0[2] = 0;
w0[1] = 0;
w0[0] = 0;
break;
case 4:
- w3[2] = amd_bytealign_S ( 0, w2[1], offset_minus_4);
- w3[1] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
- w3[0] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w2[3] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w2[2] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w2[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w2[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w1[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w1[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w1[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w1[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w2[1], offset_minus_4);
+ w3[1] = amd_bytealign (w2[1], w2[0], offset_minus_4);
+ w3[0] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w2[3] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w2[2] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w2[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w2[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w1[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w1[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w1[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w1[0] = amd_bytealign (w0[0], 0, offset_minus_4);
w0[3] = 0;
w0[2] = 0;
w0[1] = 0;
break;
case 5:
- w3[2] = amd_bytealign_S ( 0, w2[0], offset_minus_4);
- w3[1] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
- w3[0] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w2[3] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w2[2] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w2[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w2[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w1[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w1[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w1[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w2[0], offset_minus_4);
+ w3[1] = amd_bytealign (w2[0], w1[3], offset_minus_4);
+ w3[0] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w2[3] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w2[2] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w2[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w2[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w1[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w1[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w1[1] = amd_bytealign (w0[0], 0, offset_minus_4);
w1[0] = 0;
w0[3] = 0;
w0[2] = 0;
break;
case 6:
- w3[2] = amd_bytealign_S ( 0, w1[3], offset_minus_4);
- w3[1] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
- w3[0] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w2[3] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w2[2] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w2[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w2[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w1[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w1[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w1[3], offset_minus_4);
+ w3[1] = amd_bytealign (w1[3], w1[2], offset_minus_4);
+ w3[0] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w2[3] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w2[2] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w2[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w2[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w1[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w1[2] = amd_bytealign (w0[0], 0, offset_minus_4);
w1[1] = 0;
w1[0] = 0;
w0[3] = 0;
break;
case 7:
- w3[2] = amd_bytealign_S ( 0, w1[2], offset_minus_4);
- w3[1] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
- w3[0] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w2[3] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w2[2] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w2[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w2[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w1[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w1[2], offset_minus_4);
+ w3[1] = amd_bytealign (w1[2], w1[1], offset_minus_4);
+ w3[0] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w2[3] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w2[2] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w2[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w2[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w1[3] = amd_bytealign (w0[0], 0, offset_minus_4);
w1[2] = 0;
w1[1] = 0;
w1[0] = 0;
break;
case 8:
- w3[2] = amd_bytealign_S ( 0, w1[1], offset_minus_4);
- w3[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
- w3[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w2[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w2[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w2[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w2[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w1[1], offset_minus_4);
+ w3[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
+ w3[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w2[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w2[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w2[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w2[0] = amd_bytealign (w0[0], 0, offset_minus_4);
w1[3] = 0;
w1[2] = 0;
w1[1] = 0;
break;
case 9:
- w3[2] = amd_bytealign_S ( 0, w1[0], offset_minus_4);
- w3[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
- w3[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w2[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w2[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w2[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
- w2[0] = 0;
- w1[3] = 0;
+ w3[2] = amd_bytealign ( 0, w1[0], offset_minus_4);
+ w3[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
+ w3[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w2[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w2[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w2[1] = amd_bytealign (w0[0], 0, offset_minus_4);
+ w2[0] = 0;
+ w1[3] = 0;
w1[2] = 0;
w1[1] = 0;
w1[0] = 0;
break;
case 10:
- w3[2] = amd_bytealign_S ( 0, w0[3], offset_minus_4);
- w3[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
- w3[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w2[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w2[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w0[3], offset_minus_4);
+ w3[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
+ w3[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w2[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w2[2] = amd_bytealign (w0[0], 0, offset_minus_4);
w2[1] = 0;
w2[0] = 0;
w1[3] = 0;
break;
case 11:
- w3[2] = amd_bytealign_S ( 0, w0[2], offset_minus_4);
- w3[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
- w3[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w2[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w0[2], offset_minus_4);
+ w3[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
+ w3[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w2[3] = amd_bytealign (w0[0], 0, offset_minus_4);
w2[2] = 0;
w2[1] = 0;
w2[0] = 0;
break;
case 12:
- w3[2] = amd_bytealign_S ( 0, w0[1], offset_minus_4);
- w3[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
- w3[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w0[1], offset_minus_4);
+ w3[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
+ w3[0] = amd_bytealign (w0[0], 0, offset_minus_4);
w2[3] = 0;
w2[2] = 0;
w2[1] = 0;
break;
case 13:
- w3[2] = amd_bytealign_S ( 0, w0[0], offset_minus_4);
- w3[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[2] = amd_bytealign ( 0, w0[0], offset_minus_4);
+ w3[1] = amd_bytealign (w0[0], 0, offset_minus_4);
w3[0] = 0;
w2[3] = 0;
w2[2] = 0;
switch (offset / 4)
{
case 0:
- w3[1] = __byte_perm_S (w3[0], w3[1], selector);
- w3[0] = __byte_perm_S (w2[3], w3[0], selector);
- w2[3] = __byte_perm_S (w2[2], w2[3], selector);
- w2[2] = __byte_perm_S (w2[1], w2[2], selector);
- w2[1] = __byte_perm_S (w2[0], w2[1], selector);
- w2[0] = __byte_perm_S (w1[3], w2[0], selector);
- w1[3] = __byte_perm_S (w1[2], w1[3], selector);
- w1[2] = __byte_perm_S (w1[1], w1[2], selector);
- w1[1] = __byte_perm_S (w1[0], w1[1], selector);
- w1[0] = __byte_perm_S (w0[3], w1[0], selector);
- w0[3] = __byte_perm_S (w0[2], w0[3], selector);
- w0[2] = __byte_perm_S (w0[1], w0[2], selector);
- w0[1] = __byte_perm_S (w0[0], w0[1], selector);
- w0[0] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w3[0], w3[1], selector);
+ w3[0] = __byte_perm (w2[3], w3[0], selector);
+ w2[3] = __byte_perm (w2[2], w2[3], selector);
+ w2[2] = __byte_perm (w2[1], w2[2], selector);
+ w2[1] = __byte_perm (w2[0], w2[1], selector);
+ w2[0] = __byte_perm (w1[3], w2[0], selector);
+ w1[3] = __byte_perm (w1[2], w1[3], selector);
+ w1[2] = __byte_perm (w1[1], w1[2], selector);
+ w1[1] = __byte_perm (w1[0], w1[1], selector);
+ w1[0] = __byte_perm (w0[3], w1[0], selector);
+ w0[3] = __byte_perm (w0[2], w0[3], selector);
+ w0[2] = __byte_perm (w0[1], w0[2], selector);
+ w0[1] = __byte_perm (w0[0], w0[1], selector);
+ w0[0] = __byte_perm ( 0, w0[0], selector);
break;
case 1:
- w3[1] = __byte_perm_S (w2[3], w3[0], selector);
- w3[0] = __byte_perm_S (w2[2], w2[3], selector);
- w2[3] = __byte_perm_S (w2[1], w2[2], selector);
- w2[2] = __byte_perm_S (w2[0], w2[1], selector);
- w2[1] = __byte_perm_S (w1[3], w2[0], selector);
- w2[0] = __byte_perm_S (w1[2], w1[3], selector);
- w1[3] = __byte_perm_S (w1[1], w1[2], selector);
- w1[2] = __byte_perm_S (w1[0], w1[1], selector);
- w1[1] = __byte_perm_S (w0[3], w1[0], selector);
- w1[0] = __byte_perm_S (w0[2], w0[3], selector);
- w0[3] = __byte_perm_S (w0[1], w0[2], selector);
- w0[2] = __byte_perm_S (w0[0], w0[1], selector);
- w0[1] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w2[3], w3[0], selector);
+ w3[0] = __byte_perm (w2[2], w2[3], selector);
+ w2[3] = __byte_perm (w2[1], w2[2], selector);
+ w2[2] = __byte_perm (w2[0], w2[1], selector);
+ w2[1] = __byte_perm (w1[3], w2[0], selector);
+ w2[0] = __byte_perm (w1[2], w1[3], selector);
+ w1[3] = __byte_perm (w1[1], w1[2], selector);
+ w1[2] = __byte_perm (w1[0], w1[1], selector);
+ w1[1] = __byte_perm (w0[3], w1[0], selector);
+ w1[0] = __byte_perm (w0[2], w0[3], selector);
+ w0[3] = __byte_perm (w0[1], w0[2], selector);
+ w0[2] = __byte_perm (w0[0], w0[1], selector);
+ w0[1] = __byte_perm ( 0, w0[0], selector);
w0[0] = 0;
break;
case 2:
- w3[1] = __byte_perm_S (w2[2], w2[3], selector);
- w3[0] = __byte_perm_S (w2[1], w2[2], selector);
- w2[3] = __byte_perm_S (w2[0], w2[1], selector);
- w2[2] = __byte_perm_S (w1[3], w2[0], selector);
- w2[1] = __byte_perm_S (w1[2], w1[3], selector);
- w2[0] = __byte_perm_S (w1[1], w1[2], selector);
- w1[3] = __byte_perm_S (w1[0], w1[1], selector);
- w1[2] = __byte_perm_S (w0[3], w1[0], selector);
- w1[1] = __byte_perm_S (w0[2], w0[3], selector);
- w1[0] = __byte_perm_S (w0[1], w0[2], selector);
- w0[3] = __byte_perm_S (w0[0], w0[1], selector);
- w0[2] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w2[2], w2[3], selector);
+ w3[0] = __byte_perm (w2[1], w2[2], selector);
+ w2[3] = __byte_perm (w2[0], w2[1], selector);
+ w2[2] = __byte_perm (w1[3], w2[0], selector);
+ w2[1] = __byte_perm (w1[2], w1[3], selector);
+ w2[0] = __byte_perm (w1[1], w1[2], selector);
+ w1[3] = __byte_perm (w1[0], w1[1], selector);
+ w1[2] = __byte_perm (w0[3], w1[0], selector);
+ w1[1] = __byte_perm (w0[2], w0[3], selector);
+ w1[0] = __byte_perm (w0[1], w0[2], selector);
+ w0[3] = __byte_perm (w0[0], w0[1], selector);
+ w0[2] = __byte_perm ( 0, w0[0], selector);
w0[1] = 0;
w0[0] = 0;
break;
case 3:
- w3[1] = __byte_perm_S (w2[1], w2[2], selector);
- w3[0] = __byte_perm_S (w2[0], w2[1], selector);
- w2[3] = __byte_perm_S (w1[3], w2[0], selector);
- w2[2] = __byte_perm_S (w1[2], w1[3], selector);
- w2[1] = __byte_perm_S (w1[1], w1[2], selector);
- w2[0] = __byte_perm_S (w1[0], w1[1], selector);
- w1[3] = __byte_perm_S (w0[3], w1[0], selector);
- w1[2] = __byte_perm_S (w0[2], w0[3], selector);
- w1[1] = __byte_perm_S (w0[1], w0[2], selector);
- w1[0] = __byte_perm_S (w0[0], w0[1], selector);
- w0[3] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w2[1], w2[2], selector);
+ w3[0] = __byte_perm (w2[0], w2[1], selector);
+ w2[3] = __byte_perm (w1[3], w2[0], selector);
+ w2[2] = __byte_perm (w1[2], w1[3], selector);
+ w2[1] = __byte_perm (w1[1], w1[2], selector);
+ w2[0] = __byte_perm (w1[0], w1[1], selector);
+ w1[3] = __byte_perm (w0[3], w1[0], selector);
+ w1[2] = __byte_perm (w0[2], w0[3], selector);
+ w1[1] = __byte_perm (w0[1], w0[2], selector);
+ w1[0] = __byte_perm (w0[0], w0[1], selector);
+ w0[3] = __byte_perm ( 0, w0[0], selector);
w0[2] = 0;
w0[1] = 0;
w0[0] = 0;
break;
case 4:
- w3[1] = __byte_perm_S (w2[0], w2[1], selector);
- w3[0] = __byte_perm_S (w1[3], w2[0], selector);
- w2[3] = __byte_perm_S (w1[2], w1[3], selector);
- w2[2] = __byte_perm_S (w1[1], w1[2], selector);
- w2[1] = __byte_perm_S (w1[0], w1[1], selector);
- w2[0] = __byte_perm_S (w0[3], w1[0], selector);
- w1[3] = __byte_perm_S (w0[2], w0[3], selector);
- w1[2] = __byte_perm_S (w0[1], w0[2], selector);
- w1[1] = __byte_perm_S (w0[0], w0[1], selector);
- w1[0] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w2[0], w2[1], selector);
+ w3[0] = __byte_perm (w1[3], w2[0], selector);
+ w2[3] = __byte_perm (w1[2], w1[3], selector);
+ w2[2] = __byte_perm (w1[1], w1[2], selector);
+ w2[1] = __byte_perm (w1[0], w1[1], selector);
+ w2[0] = __byte_perm (w0[3], w1[0], selector);
+ w1[3] = __byte_perm (w0[2], w0[3], selector);
+ w1[2] = __byte_perm (w0[1], w0[2], selector);
+ w1[1] = __byte_perm (w0[0], w0[1], selector);
+ w1[0] = __byte_perm ( 0, w0[0], selector);
w0[3] = 0;
w0[2] = 0;
w0[1] = 0;
break;
case 5:
- w3[1] = __byte_perm_S (w1[3], w2[0], selector);
- w3[0] = __byte_perm_S (w1[2], w1[3], selector);
- w2[3] = __byte_perm_S (w1[1], w1[2], selector);
- w2[2] = __byte_perm_S (w1[0], w1[1], selector);
- w2[1] = __byte_perm_S (w0[3], w1[0], selector);
- w2[0] = __byte_perm_S (w0[2], w0[3], selector);
- w1[3] = __byte_perm_S (w0[1], w0[2], selector);
- w1[2] = __byte_perm_S (w0[0], w0[1], selector);
- w1[1] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w1[3], w2[0], selector);
+ w3[0] = __byte_perm (w1[2], w1[3], selector);
+ w2[3] = __byte_perm (w1[1], w1[2], selector);
+ w2[2] = __byte_perm (w1[0], w1[1], selector);
+ w2[1] = __byte_perm (w0[3], w1[0], selector);
+ w2[0] = __byte_perm (w0[2], w0[3], selector);
+ w1[3] = __byte_perm (w0[1], w0[2], selector);
+ w1[2] = __byte_perm (w0[0], w0[1], selector);
+ w1[1] = __byte_perm ( 0, w0[0], selector);
w1[0] = 0;
w0[3] = 0;
w0[2] = 0;
break;
case 6:
- w3[1] = __byte_perm_S (w1[2], w1[3], selector);
- w3[0] = __byte_perm_S (w1[1], w1[2], selector);
- w2[3] = __byte_perm_S (w1[0], w1[1], selector);
- w2[2] = __byte_perm_S (w0[3], w1[0], selector);
- w2[1] = __byte_perm_S (w0[2], w0[3], selector);
- w2[0] = __byte_perm_S (w0[1], w0[2], selector);
- w1[3] = __byte_perm_S (w0[0], w0[1], selector);
- w1[2] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w1[2], w1[3], selector);
+ w3[0] = __byte_perm (w1[1], w1[2], selector);
+ w2[3] = __byte_perm (w1[0], w1[1], selector);
+ w2[2] = __byte_perm (w0[3], w1[0], selector);
+ w2[1] = __byte_perm (w0[2], w0[3], selector);
+ w2[0] = __byte_perm (w0[1], w0[2], selector);
+ w1[3] = __byte_perm (w0[0], w0[1], selector);
+ w1[2] = __byte_perm ( 0, w0[0], selector);
w1[1] = 0;
w1[0] = 0;
w0[3] = 0;
break;
case 7:
- w3[1] = __byte_perm_S (w1[1], w1[2], selector);
- w3[0] = __byte_perm_S (w1[0], w1[1], selector);
- w2[3] = __byte_perm_S (w0[3], w1[0], selector);
- w2[2] = __byte_perm_S (w0[2], w0[3], selector);
- w2[1] = __byte_perm_S (w0[1], w0[2], selector);
- w2[0] = __byte_perm_S (w0[0], w0[1], selector);
- w1[3] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w1[1], w1[2], selector);
+ w3[0] = __byte_perm (w1[0], w1[1], selector);
+ w2[3] = __byte_perm (w0[3], w1[0], selector);
+ w2[2] = __byte_perm (w0[2], w0[3], selector);
+ w2[1] = __byte_perm (w0[1], w0[2], selector);
+ w2[0] = __byte_perm (w0[0], w0[1], selector);
+ w1[3] = __byte_perm ( 0, w0[0], selector);
w1[2] = 0;
w1[1] = 0;
w1[0] = 0;
break;
case 8:
- w3[1] = __byte_perm_S (w1[0], w1[1], selector);
- w3[0] = __byte_perm_S (w0[3], w1[0], selector);
- w2[3] = __byte_perm_S (w0[2], w0[3], selector);
- w2[2] = __byte_perm_S (w0[1], w0[2], selector);
- w2[1] = __byte_perm_S (w0[0], w0[1], selector);
- w2[0] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w1[0], w1[1], selector);
+ w3[0] = __byte_perm (w0[3], w1[0], selector);
+ w2[3] = __byte_perm (w0[2], w0[3], selector);
+ w2[2] = __byte_perm (w0[1], w0[2], selector);
+ w2[1] = __byte_perm (w0[0], w0[1], selector);
+ w2[0] = __byte_perm ( 0, w0[0], selector);
w1[3] = 0;
w1[2] = 0;
w1[1] = 0;
break;
case 9:
- w3[1] = __byte_perm_S (w0[3], w1[0], selector);
- w3[0] = __byte_perm_S (w0[2], w0[3], selector);
- w2[3] = __byte_perm_S (w0[1], w0[2], selector);
- w2[2] = __byte_perm_S (w0[0], w0[1], selector);
- w2[1] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w0[3], w1[0], selector);
+ w3[0] = __byte_perm (w0[2], w0[3], selector);
+ w2[3] = __byte_perm (w0[1], w0[2], selector);
+ w2[2] = __byte_perm (w0[0], w0[1], selector);
+ w2[1] = __byte_perm ( 0, w0[0], selector);
w2[0] = 0;
w1[3] = 0;
w1[2] = 0;
break;
case 10:
- w3[1] = __byte_perm_S (w0[2], w0[3], selector);
- w3[0] = __byte_perm_S (w0[1], w0[2], selector);
- w2[3] = __byte_perm_S (w0[0], w0[1], selector);
- w2[2] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w0[2], w0[3], selector);
+ w3[0] = __byte_perm (w0[1], w0[2], selector);
+ w2[3] = __byte_perm (w0[0], w0[1], selector);
+ w2[2] = __byte_perm ( 0, w0[0], selector);
w2[1] = 0;
w2[0] = 0;
w1[3] = 0;
break;
case 11:
- w3[1] = __byte_perm_S (w0[1], w0[2], selector);
- w3[0] = __byte_perm_S (w0[0], w0[1], selector);
- w2[3] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w0[1], w0[2], selector);
+ w3[0] = __byte_perm (w0[0], w0[1], selector);
+ w2[3] = __byte_perm ( 0, w0[0], selector);
w2[2] = 0;
w2[1] = 0;
w2[0] = 0;
break;
case 12:
- w3[1] = __byte_perm_S (w0[0], w0[1], selector);
- w3[0] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm (w0[0], w0[1], selector);
+ w3[0] = __byte_perm ( 0, w0[0], selector);
w2[3] = 0;
w2[2] = 0;
w2[1] = 0;
break;
case 13:
- w3[1] = __byte_perm_S ( 0, w0[0], selector);
+ w3[1] = __byte_perm ( 0, w0[0], selector);
w3[0] = 0;
w2[3] = 0;
w2[2] = 0;
#endif
}
-static void switch_buffer_by_offset_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+inline void switch_buffer_by_offset_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
{
#if defined IS_AMD || defined IS_GENERIC
switch (offset / 4)
{
case 0:
- w3[2] = amd_bytealign_S (w3[1], 0, offset);
- w3[1] = amd_bytealign_S (w3[0], w3[1], offset);
- w3[0] = amd_bytealign_S (w2[3], w3[0], offset);
- w2[3] = amd_bytealign_S (w2[2], w2[3], offset);
- w2[2] = amd_bytealign_S (w2[1], w2[2], offset);
- w2[1] = amd_bytealign_S (w2[0], w2[1], offset);
- w2[0] = amd_bytealign_S (w1[3], w2[0], offset);
- w1[3] = amd_bytealign_S (w1[2], w1[3], offset);
- w1[2] = amd_bytealign_S (w1[1], w1[2], offset);
- w1[1] = amd_bytealign_S (w1[0], w1[1], offset);
- w1[0] = amd_bytealign_S (w0[3], w1[0], offset);
- w0[3] = amd_bytealign_S (w0[2], w0[3], offset);
- w0[2] = amd_bytealign_S (w0[1], w0[2], offset);
- w0[1] = amd_bytealign_S (w0[0], w0[1], offset);
- w0[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w3[1], 0, offset);
+ w3[1] = amd_bytealign (w3[0], w3[1], offset);
+ w3[0] = amd_bytealign (w2[3], w3[0], offset);
+ w2[3] = amd_bytealign (w2[2], w2[3], offset);
+ w2[2] = amd_bytealign (w2[1], w2[2], offset);
+ w2[1] = amd_bytealign (w2[0], w2[1], offset);
+ w2[0] = amd_bytealign (w1[3], w2[0], offset);
+ w1[3] = amd_bytealign (w1[2], w1[3], offset);
+ w1[2] = amd_bytealign (w1[1], w1[2], offset);
+ w1[1] = amd_bytealign (w1[0], w1[1], offset);
+ w1[0] = amd_bytealign (w0[3], w1[0], offset);
+ w0[3] = amd_bytealign (w0[2], w0[3], offset);
+ w0[2] = amd_bytealign (w0[1], w0[2], offset);
+ w0[1] = amd_bytealign (w0[0], w0[1], offset);
+ w0[0] = amd_bytealign ( 0, w0[0], offset);
break;
case 1:
- w3[2] = amd_bytealign_S (w3[0], 0, offset);
- w3[1] = amd_bytealign_S (w2[3], w3[0], offset);
- w3[0] = amd_bytealign_S (w2[2], w2[3], offset);
- w2[3] = amd_bytealign_S (w2[1], w2[2], offset);
- w2[2] = amd_bytealign_S (w2[0], w2[1], offset);
- w2[1] = amd_bytealign_S (w1[3], w2[0], offset);
- w2[0] = amd_bytealign_S (w1[2], w1[3], offset);
- w1[3] = amd_bytealign_S (w1[1], w1[2], offset);
- w1[2] = amd_bytealign_S (w1[0], w1[1], offset);
- w1[1] = amd_bytealign_S (w0[3], w1[0], offset);
- w1[0] = amd_bytealign_S (w0[2], w0[3], offset);
- w0[3] = amd_bytealign_S (w0[1], w0[2], offset);
- w0[2] = amd_bytealign_S (w0[0], w0[1], offset);
- w0[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w3[0], 0, offset);
+ w3[1] = amd_bytealign (w2[3], w3[0], offset);
+ w3[0] = amd_bytealign (w2[2], w2[3], offset);
+ w2[3] = amd_bytealign (w2[1], w2[2], offset);
+ w2[2] = amd_bytealign (w2[0], w2[1], offset);
+ w2[1] = amd_bytealign (w1[3], w2[0], offset);
+ w2[0] = amd_bytealign (w1[2], w1[3], offset);
+ w1[3] = amd_bytealign (w1[1], w1[2], offset);
+ w1[2] = amd_bytealign (w1[0], w1[1], offset);
+ w1[1] = amd_bytealign (w0[3], w1[0], offset);
+ w1[0] = amd_bytealign (w0[2], w0[3], offset);
+ w0[3] = amd_bytealign (w0[1], w0[2], offset);
+ w0[2] = amd_bytealign (w0[0], w0[1], offset);
+ w0[1] = amd_bytealign ( 0, w0[0], offset);
w0[0] = 0;
break;
case 2:
- w3[2] = amd_bytealign_S (w2[3], 0, offset);
- w3[1] = amd_bytealign_S (w2[2], w2[3], offset);
- w3[0] = amd_bytealign_S (w2[1], w2[2], offset);
- w2[3] = amd_bytealign_S (w2[0], w2[1], offset);
- w2[2] = amd_bytealign_S (w1[3], w2[0], offset);
- w2[1] = amd_bytealign_S (w1[2], w1[3], offset);
- w2[0] = amd_bytealign_S (w1[1], w1[2], offset);
- w1[3] = amd_bytealign_S (w1[0], w1[1], offset);
- w1[2] = amd_bytealign_S (w0[3], w1[0], offset);
- w1[1] = amd_bytealign_S (w0[2], w0[3], offset);
- w1[0] = amd_bytealign_S (w0[1], w0[2], offset);
- w0[3] = amd_bytealign_S (w0[0], w0[1], offset);
- w0[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w2[3], 0, offset);
+ w3[1] = amd_bytealign (w2[2], w2[3], offset);
+ w3[0] = amd_bytealign (w2[1], w2[2], offset);
+ w2[3] = amd_bytealign (w2[0], w2[1], offset);
+ w2[2] = amd_bytealign (w1[3], w2[0], offset);
+ w2[1] = amd_bytealign (w1[2], w1[3], offset);
+ w2[0] = amd_bytealign (w1[1], w1[2], offset);
+ w1[3] = amd_bytealign (w1[0], w1[1], offset);
+ w1[2] = amd_bytealign (w0[3], w1[0], offset);
+ w1[1] = amd_bytealign (w0[2], w0[3], offset);
+ w1[0] = amd_bytealign (w0[1], w0[2], offset);
+ w0[3] = amd_bytealign (w0[0], w0[1], offset);
+ w0[2] = amd_bytealign ( 0, w0[0], offset);
w0[1] = 0;
w0[0] = 0;
break;
case 3:
- w3[2] = amd_bytealign_S (w2[2], 0, offset);
- w3[1] = amd_bytealign_S (w2[1], w2[2], offset);
- w3[0] = amd_bytealign_S (w2[0], w2[1], offset);
- w2[3] = amd_bytealign_S (w1[3], w2[0], offset);
- w2[2] = amd_bytealign_S (w1[2], w1[3], offset);
- w2[1] = amd_bytealign_S (w1[1], w1[2], offset);
- w2[0] = amd_bytealign_S (w1[0], w1[1], offset);
- w1[3] = amd_bytealign_S (w0[3], w1[0], offset);
- w1[2] = amd_bytealign_S (w0[2], w0[3], offset);
- w1[1] = amd_bytealign_S (w0[1], w0[2], offset);
- w1[0] = amd_bytealign_S (w0[0], w0[1], offset);
- w0[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w2[2], 0, offset);
+ w3[1] = amd_bytealign (w2[1], w2[2], offset);
+ w3[0] = amd_bytealign (w2[0], w2[1], offset);
+ w2[3] = amd_bytealign (w1[3], w2[0], offset);
+ w2[2] = amd_bytealign (w1[2], w1[3], offset);
+ w2[1] = amd_bytealign (w1[1], w1[2], offset);
+ w2[0] = amd_bytealign (w1[0], w1[1], offset);
+ w1[3] = amd_bytealign (w0[3], w1[0], offset);
+ w1[2] = amd_bytealign (w0[2], w0[3], offset);
+ w1[1] = amd_bytealign (w0[1], w0[2], offset);
+ w1[0] = amd_bytealign (w0[0], w0[1], offset);
+ w0[3] = amd_bytealign ( 0, w0[0], offset);
w0[2] = 0;
w0[1] = 0;
w0[0] = 0;
break;
case 4:
- w3[2] = amd_bytealign_S (w2[1], 0, offset);
- w3[1] = amd_bytealign_S (w2[0], w2[1], offset);
- w3[0] = amd_bytealign_S (w1[3], w2[0], offset);
- w2[3] = amd_bytealign_S (w1[2], w1[3], offset);
- w2[2] = amd_bytealign_S (w1[1], w1[2], offset);
- w2[1] = amd_bytealign_S (w1[0], w1[1], offset);
- w2[0] = amd_bytealign_S (w0[3], w1[0], offset);
- w1[3] = amd_bytealign_S (w0[2], w0[3], offset);
- w1[2] = amd_bytealign_S (w0[1], w0[2], offset);
- w1[1] = amd_bytealign_S (w0[0], w0[1], offset);
- w1[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w2[1], 0, offset);
+ w3[1] = amd_bytealign (w2[0], w2[1], offset);
+ w3[0] = amd_bytealign (w1[3], w2[0], offset);
+ w2[3] = amd_bytealign (w1[2], w1[3], offset);
+ w2[2] = amd_bytealign (w1[1], w1[2], offset);
+ w2[1] = amd_bytealign (w1[0], w1[1], offset);
+ w2[0] = amd_bytealign (w0[3], w1[0], offset);
+ w1[3] = amd_bytealign (w0[2], w0[3], offset);
+ w1[2] = amd_bytealign (w0[1], w0[2], offset);
+ w1[1] = amd_bytealign (w0[0], w0[1], offset);
+ w1[0] = amd_bytealign ( 0, w0[0], offset);
w0[3] = 0;
w0[2] = 0;
w0[1] = 0;
break;
case 5:
- w3[2] = amd_bytealign_S (w2[0], 0, offset);
- w3[1] = amd_bytealign_S (w1[3], w2[0], offset);
- w3[0] = amd_bytealign_S (w1[2], w1[3], offset);
- w2[3] = amd_bytealign_S (w1[1], w1[2], offset);
- w2[2] = amd_bytealign_S (w1[0], w1[1], offset);
- w2[1] = amd_bytealign_S (w0[3], w1[0], offset);
- w2[0] = amd_bytealign_S (w0[2], w0[3], offset);
- w1[3] = amd_bytealign_S (w0[1], w0[2], offset);
- w1[2] = amd_bytealign_S (w0[0], w0[1], offset);
- w1[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w2[0], 0, offset);
+ w3[1] = amd_bytealign (w1[3], w2[0], offset);
+ w3[0] = amd_bytealign (w1[2], w1[3], offset);
+ w2[3] = amd_bytealign (w1[1], w1[2], offset);
+ w2[2] = amd_bytealign (w1[0], w1[1], offset);
+ w2[1] = amd_bytealign (w0[3], w1[0], offset);
+ w2[0] = amd_bytealign (w0[2], w0[3], offset);
+ w1[3] = amd_bytealign (w0[1], w0[2], offset);
+ w1[2] = amd_bytealign (w0[0], w0[1], offset);
+ w1[1] = amd_bytealign ( 0, w0[0], offset);
w1[0] = 0;
w0[3] = 0;
w0[2] = 0;
break;
case 6:
- w3[2] = amd_bytealign_S (w1[3], 0, offset);
- w3[1] = amd_bytealign_S (w1[2], w1[3], offset);
- w3[0] = amd_bytealign_S (w1[1], w1[2], offset);
- w2[3] = amd_bytealign_S (w1[0], w1[1], offset);
- w2[2] = amd_bytealign_S (w0[3], w1[0], offset);
- w2[1] = amd_bytealign_S (w0[2], w0[3], offset);
- w2[0] = amd_bytealign_S (w0[1], w0[2], offset);
- w1[3] = amd_bytealign_S (w0[0], w0[1], offset);
- w1[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w1[3], 0, offset);
+ w3[1] = amd_bytealign (w1[2], w1[3], offset);
+ w3[0] = amd_bytealign (w1[1], w1[2], offset);
+ w2[3] = amd_bytealign (w1[0], w1[1], offset);
+ w2[2] = amd_bytealign (w0[3], w1[0], offset);
+ w2[1] = amd_bytealign (w0[2], w0[3], offset);
+ w2[0] = amd_bytealign (w0[1], w0[2], offset);
+ w1[3] = amd_bytealign (w0[0], w0[1], offset);
+ w1[2] = amd_bytealign ( 0, w0[0], offset);
w1[1] = 0;
w1[0] = 0;
w0[3] = 0;
break;
case 7:
- w3[2] = amd_bytealign_S (w1[2], 0, offset);
- w3[1] = amd_bytealign_S (w1[1], w1[2], offset);
- w3[0] = amd_bytealign_S (w1[0], w1[1], offset);
- w2[3] = amd_bytealign_S (w0[3], w1[0], offset);
- w2[2] = amd_bytealign_S (w0[2], w0[3], offset);
- w2[1] = amd_bytealign_S (w0[1], w0[2], offset);
- w2[0] = amd_bytealign_S (w0[0], w0[1], offset);
- w1[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w1[2], 0, offset);
+ w3[1] = amd_bytealign (w1[1], w1[2], offset);
+ w3[0] = amd_bytealign (w1[0], w1[1], offset);
+ w2[3] = amd_bytealign (w0[3], w1[0], offset);
+ w2[2] = amd_bytealign (w0[2], w0[3], offset);
+ w2[1] = amd_bytealign (w0[1], w0[2], offset);
+ w2[0] = amd_bytealign (w0[0], w0[1], offset);
+ w1[3] = amd_bytealign ( 0, w0[0], offset);
w1[2] = 0;
w1[1] = 0;
w1[0] = 0;
break;
case 8:
- w3[2] = amd_bytealign_S (w1[1], 0, offset);
- w3[1] = amd_bytealign_S (w1[0], w1[1], offset);
- w3[0] = amd_bytealign_S (w0[3], w1[0], offset);
- w2[3] = amd_bytealign_S (w0[2], w0[3], offset);
- w2[2] = amd_bytealign_S (w0[1], w0[2], offset);
- w2[1] = amd_bytealign_S (w0[0], w0[1], offset);
- w2[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w1[1], 0, offset);
+ w3[1] = amd_bytealign (w1[0], w1[1], offset);
+ w3[0] = amd_bytealign (w0[3], w1[0], offset);
+ w2[3] = amd_bytealign (w0[2], w0[3], offset);
+ w2[2] = amd_bytealign (w0[1], w0[2], offset);
+ w2[1] = amd_bytealign (w0[0], w0[1], offset);
+ w2[0] = amd_bytealign ( 0, w0[0], offset);
w1[3] = 0;
w1[2] = 0;
w1[1] = 0;
break;
case 9:
- w3[2] = amd_bytealign_S (w1[0], 0, offset);
- w3[1] = amd_bytealign_S (w0[3], w1[0], offset);
- w3[0] = amd_bytealign_S (w0[2], w0[3], offset);
- w2[3] = amd_bytealign_S (w0[1], w0[2], offset);
- w2[2] = amd_bytealign_S (w0[0], w0[1], offset);
- w2[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w1[0], 0, offset);
+ w3[1] = amd_bytealign (w0[3], w1[0], offset);
+ w3[0] = amd_bytealign (w0[2], w0[3], offset);
+ w2[3] = amd_bytealign (w0[1], w0[2], offset);
+ w2[2] = amd_bytealign (w0[0], w0[1], offset);
+ w2[1] = amd_bytealign ( 0, w0[0], offset);
w2[0] = 0;
w1[3] = 0;
w1[2] = 0;
break;
case 10:
- w3[2] = amd_bytealign_S (w0[3], 0, offset);
- w3[1] = amd_bytealign_S (w0[2], w0[3], offset);
- w3[0] = amd_bytealign_S (w0[1], w0[2], offset);
- w2[3] = amd_bytealign_S (w0[0], w0[1], offset);
- w2[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w0[3], 0, offset);
+ w3[1] = amd_bytealign (w0[2], w0[3], offset);
+ w3[0] = amd_bytealign (w0[1], w0[2], offset);
+ w2[3] = amd_bytealign (w0[0], w0[1], offset);
+ w2[2] = amd_bytealign ( 0, w0[0], offset);
w2[1] = 0;
w2[0] = 0;
w1[3] = 0;
break;
case 11:
- w3[2] = amd_bytealign_S (w0[2], 0, offset);
- w3[1] = amd_bytealign_S (w0[1], w0[2], offset);
- w3[0] = amd_bytealign_S (w0[0], w0[1], offset);
- w2[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w0[2], 0, offset);
+ w3[1] = amd_bytealign (w0[1], w0[2], offset);
+ w3[0] = amd_bytealign (w0[0], w0[1], offset);
+ w2[3] = amd_bytealign ( 0, w0[0], offset);
w2[2] = 0;
w2[1] = 0;
w2[0] = 0;
break;
case 12:
- w3[2] = amd_bytealign_S (w0[1], 0, offset);
- w3[1] = amd_bytealign_S (w0[0], w0[1], offset);
- w3[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w0[1], 0, offset);
+ w3[1] = amd_bytealign (w0[0], w0[1], offset);
+ w3[0] = amd_bytealign ( 0, w0[0], offset);
w2[3] = 0;
w2[2] = 0;
w2[1] = 0;
break;
case 13:
- w3[2] = amd_bytealign_S (w0[0], 0, offset);
- w3[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[2] = amd_bytealign (w0[0], 0, offset);
+ w3[1] = amd_bytealign ( 0, w0[0], offset);
w3[0] = 0;
w2[3] = 0;
w2[2] = 0;
switch (offset / 4)
{
case 0:
- w3[1] = __byte_perm_S (w3[1], w3[0], selector);
- w3[0] = __byte_perm_S (w3[0], w2[3], selector);
- w2[3] = __byte_perm_S (w2[3], w2[2], selector);
- w2[2] = __byte_perm_S (w2[2], w2[1], selector);
- w2[1] = __byte_perm_S (w2[1], w2[0], selector);
- w2[0] = __byte_perm_S (w2[0], w1[3], selector);
- w1[3] = __byte_perm_S (w1[3], w1[2], selector);
- w1[2] = __byte_perm_S (w1[2], w1[1], selector);
- w1[1] = __byte_perm_S (w1[1], w1[0], selector);
- w1[0] = __byte_perm_S (w1[0], w0[3], selector);
- w0[3] = __byte_perm_S (w0[3], w0[2], selector);
- w0[2] = __byte_perm_S (w0[2], w0[1], selector);
- w0[1] = __byte_perm_S (w0[1], w0[0], selector);
- w0[0] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w3[1], w3[0], selector);
+ w3[0] = __byte_perm (w3[0], w2[3], selector);
+ w2[3] = __byte_perm (w2[3], w2[2], selector);
+ w2[2] = __byte_perm (w2[2], w2[1], selector);
+ w2[1] = __byte_perm (w2[1], w2[0], selector);
+ w2[0] = __byte_perm (w2[0], w1[3], selector);
+ w1[3] = __byte_perm (w1[3], w1[2], selector);
+ w1[2] = __byte_perm (w1[2], w1[1], selector);
+ w1[1] = __byte_perm (w1[1], w1[0], selector);
+ w1[0] = __byte_perm (w1[0], w0[3], selector);
+ w0[3] = __byte_perm (w0[3], w0[2], selector);
+ w0[2] = __byte_perm (w0[2], w0[1], selector);
+ w0[1] = __byte_perm (w0[1], w0[0], selector);
+ w0[0] = __byte_perm (w0[0], 0, selector);
break;
case 1:
- w3[1] = __byte_perm_S (w3[0], w2[3], selector);
- w3[0] = __byte_perm_S (w2[3], w2[2], selector);
- w2[3] = __byte_perm_S (w2[2], w2[1], selector);
- w2[2] = __byte_perm_S (w2[1], w2[0], selector);
- w2[1] = __byte_perm_S (w2[0], w1[3], selector);
- w2[0] = __byte_perm_S (w1[3], w1[2], selector);
- w1[3] = __byte_perm_S (w1[2], w1[1], selector);
- w1[2] = __byte_perm_S (w1[1], w1[0], selector);
- w1[1] = __byte_perm_S (w1[0], w0[3], selector);
- w1[0] = __byte_perm_S (w0[3], w0[2], selector);
- w0[3] = __byte_perm_S (w0[2], w0[1], selector);
- w0[2] = __byte_perm_S (w0[1], w0[0], selector);
- w0[1] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w3[0], w2[3], selector);
+ w3[0] = __byte_perm (w2[3], w2[2], selector);
+ w2[3] = __byte_perm (w2[2], w2[1], selector);
+ w2[2] = __byte_perm (w2[1], w2[0], selector);
+ w2[1] = __byte_perm (w2[0], w1[3], selector);
+ w2[0] = __byte_perm (w1[3], w1[2], selector);
+ w1[3] = __byte_perm (w1[2], w1[1], selector);
+ w1[2] = __byte_perm (w1[1], w1[0], selector);
+ w1[1] = __byte_perm (w1[0], w0[3], selector);
+ w1[0] = __byte_perm (w0[3], w0[2], selector);
+ w0[3] = __byte_perm (w0[2], w0[1], selector);
+ w0[2] = __byte_perm (w0[1], w0[0], selector);
+ w0[1] = __byte_perm (w0[0], 0, selector);
w0[0] = 0;
break;
case 2:
- w3[1] = __byte_perm_S (w2[3], w2[2], selector);
- w3[0] = __byte_perm_S (w2[2], w2[1], selector);
- w2[3] = __byte_perm_S (w2[1], w2[0], selector);
- w2[2] = __byte_perm_S (w2[0], w1[3], selector);
- w2[1] = __byte_perm_S (w1[3], w1[2], selector);
- w2[0] = __byte_perm_S (w1[2], w1[1], selector);
- w1[3] = __byte_perm_S (w1[1], w1[0], selector);
- w1[2] = __byte_perm_S (w1[0], w0[3], selector);
- w1[1] = __byte_perm_S (w0[3], w0[2], selector);
- w1[0] = __byte_perm_S (w0[2], w0[1], selector);
- w0[3] = __byte_perm_S (w0[1], w0[0], selector);
- w0[2] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w2[3], w2[2], selector);
+ w3[0] = __byte_perm (w2[2], w2[1], selector);
+ w2[3] = __byte_perm (w2[1], w2[0], selector);
+ w2[2] = __byte_perm (w2[0], w1[3], selector);
+ w2[1] = __byte_perm (w1[3], w1[2], selector);
+ w2[0] = __byte_perm (w1[2], w1[1], selector);
+ w1[3] = __byte_perm (w1[1], w1[0], selector);
+ w1[2] = __byte_perm (w1[0], w0[3], selector);
+ w1[1] = __byte_perm (w0[3], w0[2], selector);
+ w1[0] = __byte_perm (w0[2], w0[1], selector);
+ w0[3] = __byte_perm (w0[1], w0[0], selector);
+ w0[2] = __byte_perm (w0[0], 0, selector);
w0[1] = 0;
w0[0] = 0;
break;
case 3:
- w3[1] = __byte_perm_S (w2[2], w2[1], selector);
- w3[0] = __byte_perm_S (w2[1], w2[0], selector);
- w2[3] = __byte_perm_S (w2[0], w1[3], selector);
- w2[2] = __byte_perm_S (w1[3], w1[2], selector);
- w2[1] = __byte_perm_S (w1[2], w1[1], selector);
- w2[0] = __byte_perm_S (w1[1], w1[0], selector);
- w1[3] = __byte_perm_S (w1[0], w0[3], selector);
- w1[2] = __byte_perm_S (w0[3], w0[2], selector);
- w1[1] = __byte_perm_S (w0[2], w0[1], selector);
- w1[0] = __byte_perm_S (w0[1], w0[0], selector);
- w0[3] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w2[2], w2[1], selector);
+ w3[0] = __byte_perm (w2[1], w2[0], selector);
+ w2[3] = __byte_perm (w2[0], w1[3], selector);
+ w2[2] = __byte_perm (w1[3], w1[2], selector);
+ w2[1] = __byte_perm (w1[2], w1[1], selector);
+ w2[0] = __byte_perm (w1[1], w1[0], selector);
+ w1[3] = __byte_perm (w1[0], w0[3], selector);
+ w1[2] = __byte_perm (w0[3], w0[2], selector);
+ w1[1] = __byte_perm (w0[2], w0[1], selector);
+ w1[0] = __byte_perm (w0[1], w0[0], selector);
+ w0[3] = __byte_perm (w0[0], 0, selector);
w0[2] = 0;
w0[1] = 0;
w0[0] = 0;
break;
case 4:
- w3[1] = __byte_perm_S (w2[1], w2[0], selector);
- w3[0] = __byte_perm_S (w2[0], w1[3], selector);
- w2[3] = __byte_perm_S (w1[3], w1[2], selector);
- w2[2] = __byte_perm_S (w1[2], w1[1], selector);
- w2[1] = __byte_perm_S (w1[1], w1[0], selector);
- w2[0] = __byte_perm_S (w1[0], w0[3], selector);
- w1[3] = __byte_perm_S (w0[3], w0[2], selector);
- w1[2] = __byte_perm_S (w0[2], w0[1], selector);
- w1[1] = __byte_perm_S (w0[1], w0[0], selector);
- w1[0] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w2[1], w2[0], selector);
+ w3[0] = __byte_perm (w2[0], w1[3], selector);
+ w2[3] = __byte_perm (w1[3], w1[2], selector);
+ w2[2] = __byte_perm (w1[2], w1[1], selector);
+ w2[1] = __byte_perm (w1[1], w1[0], selector);
+ w2[0] = __byte_perm (w1[0], w0[3], selector);
+ w1[3] = __byte_perm (w0[3], w0[2], selector);
+ w1[2] = __byte_perm (w0[2], w0[1], selector);
+ w1[1] = __byte_perm (w0[1], w0[0], selector);
+ w1[0] = __byte_perm (w0[0], 0, selector);
w0[3] = 0;
w0[2] = 0;
w0[1] = 0;
break;
case 5:
- w3[1] = __byte_perm_S (w2[0], w1[3], selector);
- w3[0] = __byte_perm_S (w1[3], w1[2], selector);
- w2[3] = __byte_perm_S (w1[2], w1[1], selector);
- w2[2] = __byte_perm_S (w1[1], w1[0], selector);
- w2[1] = __byte_perm_S (w1[0], w0[3], selector);
- w2[0] = __byte_perm_S (w0[3], w0[2], selector);
- w1[3] = __byte_perm_S (w0[2], w0[1], selector);
- w1[2] = __byte_perm_S (w0[1], w0[0], selector);
- w1[1] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w2[0], w1[3], selector);
+ w3[0] = __byte_perm (w1[3], w1[2], selector);
+ w2[3] = __byte_perm (w1[2], w1[1], selector);
+ w2[2] = __byte_perm (w1[1], w1[0], selector);
+ w2[1] = __byte_perm (w1[0], w0[3], selector);
+ w2[0] = __byte_perm (w0[3], w0[2], selector);
+ w1[3] = __byte_perm (w0[2], w0[1], selector);
+ w1[2] = __byte_perm (w0[1], w0[0], selector);
+ w1[1] = __byte_perm (w0[0], 0, selector);
w1[0] = 0;
w0[3] = 0;
w0[2] = 0;
break;
case 6:
- w3[1] = __byte_perm_S (w1[3], w1[2], selector);
- w3[0] = __byte_perm_S (w1[2], w1[1], selector);
- w2[3] = __byte_perm_S (w1[1], w1[0], selector);
- w2[2] = __byte_perm_S (w1[0], w0[3], selector);
- w2[1] = __byte_perm_S (w0[3], w0[2], selector);
- w2[0] = __byte_perm_S (w0[2], w0[1], selector);
- w1[3] = __byte_perm_S (w0[1], w0[0], selector);
- w1[2] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w1[3], w1[2], selector);
+ w3[0] = __byte_perm (w1[2], w1[1], selector);
+ w2[3] = __byte_perm (w1[1], w1[0], selector);
+ w2[2] = __byte_perm (w1[0], w0[3], selector);
+ w2[1] = __byte_perm (w0[3], w0[2], selector);
+ w2[0] = __byte_perm (w0[2], w0[1], selector);
+ w1[3] = __byte_perm (w0[1], w0[0], selector);
+ w1[2] = __byte_perm (w0[0], 0, selector);
w1[1] = 0;
w1[0] = 0;
w0[3] = 0;
break;
case 7:
- w3[1] = __byte_perm_S (w1[2], w1[1], selector);
- w3[0] = __byte_perm_S (w1[1], w1[0], selector);
- w2[3] = __byte_perm_S (w1[0], w0[3], selector);
- w2[2] = __byte_perm_S (w0[3], w0[2], selector);
- w2[1] = __byte_perm_S (w0[2], w0[1], selector);
- w2[0] = __byte_perm_S (w0[1], w0[0], selector);
- w1[3] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w1[2], w1[1], selector);
+ w3[0] = __byte_perm (w1[1], w1[0], selector);
+ w2[3] = __byte_perm (w1[0], w0[3], selector);
+ w2[2] = __byte_perm (w0[3], w0[2], selector);
+ w2[1] = __byte_perm (w0[2], w0[1], selector);
+ w2[0] = __byte_perm (w0[1], w0[0], selector);
+ w1[3] = __byte_perm (w0[0], 0, selector);
w1[2] = 0;
w1[1] = 0;
w1[0] = 0;
break;
case 8:
- w3[1] = __byte_perm_S (w1[1], w1[0], selector);
- w3[0] = __byte_perm_S (w1[0], w0[3], selector);
- w2[3] = __byte_perm_S (w0[3], w0[2], selector);
- w2[2] = __byte_perm_S (w0[2], w0[1], selector);
- w2[1] = __byte_perm_S (w0[1], w0[0], selector);
- w2[0] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w1[1], w1[0], selector);
+ w3[0] = __byte_perm (w1[0], w0[3], selector);
+ w2[3] = __byte_perm (w0[3], w0[2], selector);
+ w2[2] = __byte_perm (w0[2], w0[1], selector);
+ w2[1] = __byte_perm (w0[1], w0[0], selector);
+ w2[0] = __byte_perm (w0[0], 0, selector);
w1[3] = 0;
w1[2] = 0;
w1[1] = 0;
break;
case 9:
- w3[1] = __byte_perm_S (w1[0], w0[3], selector);
- w3[0] = __byte_perm_S (w0[3], w0[2], selector);
- w2[3] = __byte_perm_S (w0[2], w0[1], selector);
- w2[2] = __byte_perm_S (w0[1], w0[0], selector);
- w2[1] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w1[0], w0[3], selector);
+ w3[0] = __byte_perm (w0[3], w0[2], selector);
+ w2[3] = __byte_perm (w0[2], w0[1], selector);
+ w2[2] = __byte_perm (w0[1], w0[0], selector);
+ w2[1] = __byte_perm (w0[0], 0, selector);
w2[0] = 0;
w1[3] = 0;
w1[2] = 0;
break;
case 10:
- w3[1] = __byte_perm_S (w0[3], w0[2], selector);
- w3[0] = __byte_perm_S (w0[2], w0[1], selector);
- w2[3] = __byte_perm_S (w0[1], w0[0], selector);
- w2[2] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w0[3], w0[2], selector);
+ w3[0] = __byte_perm (w0[2], w0[1], selector);
+ w2[3] = __byte_perm (w0[1], w0[0], selector);
+ w2[2] = __byte_perm (w0[0], 0, selector);
w2[1] = 0;
w2[0] = 0;
w1[3] = 0;
break;
case 11:
- w3[1] = __byte_perm_S (w0[2], w0[1], selector);
- w3[0] = __byte_perm_S (w0[1], w0[0], selector);
- w2[3] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w0[2], w0[1], selector);
+ w3[0] = __byte_perm (w0[1], w0[0], selector);
+ w2[3] = __byte_perm (w0[0], 0, selector);
w2[2] = 0;
w2[1] = 0;
w2[0] = 0;
break;
case 12:
- w3[1] = __byte_perm_S (w0[1], w0[0], selector);
- w3[0] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w0[1], w0[0], selector);
+ w3[0] = __byte_perm (w0[0], 0, selector);
w2[3] = 0;
w2[2] = 0;
w2[1] = 0;
break;
case 13:
- w3[1] = __byte_perm_S (w0[0], 0, selector);
+ w3[1] = __byte_perm (w0[0], 0, selector);
w3[0] = 0;
w2[3] = 0;
w2[2] = 0;
#endif
}
-static void switch_buffer_by_offset (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
+inline void overwrite_at_le (u32x sw[16], const u32x w0, const u32 salt_len)
{
- #if defined IS_AMD || defined IS_GENERIC
- const int offset_mod_4 = offset & 3;
+ #if defined cl_amd_media_ops
+ switch (salt_len)
+ {
+ case 0: sw[0] = w0;
+ break;
+ case 1: sw[0] = amd_bytealign (w0, sw[0] << 24, 3);
+ sw[1] = amd_bytealign (sw[1] >> 8, w0, 3);
+ break;
+ case 2: sw[0] = amd_bytealign (w0, sw[0] << 16, 2);
+ sw[1] = amd_bytealign (sw[1] >> 16, w0, 2);
+ break;
+ case 3: sw[0] = amd_bytealign (w0, sw[0] << 8, 1);
+ sw[1] = amd_bytealign (sw[1] >> 24, w0, 1);
+ break;
+ case 4: sw[1] = w0;
+ break;
+ case 5: sw[1] = amd_bytealign (w0, sw[1] << 24, 3);
+ sw[2] = amd_bytealign (sw[2] >> 8, w0, 3);
+ break;
+ case 6: sw[1] = amd_bytealign (w0, sw[1] << 16, 2);
+ sw[2] = amd_bytealign (sw[2] >> 16, w0, 2);
+ break;
+ case 7: sw[1] = amd_bytealign (w0, sw[1] << 8, 1);
+ sw[2] = amd_bytealign (sw[2] >> 24, w0, 1);
+ break;
+ case 8: sw[2] = w0;
+ break;
+ case 9: sw[2] = amd_bytealign (w0, sw[2] << 24, 3);
+ sw[3] = amd_bytealign (sw[3] >> 8, w0, 3);
+ break;
+ case 10: sw[2] = amd_bytealign (w0, sw[2] << 16, 2);
+ sw[3] = amd_bytealign (sw[3] >> 16, w0, 2);
+ break;
+ case 11: sw[2] = amd_bytealign (w0, sw[2] << 8, 1);
+ sw[3] = amd_bytealign (sw[3] >> 24, w0, 1);
+ break;
+ case 12: sw[3] = w0;
+ break;
+ case 13: sw[3] = amd_bytealign (w0, sw[3] << 24, 3);
+ sw[4] = amd_bytealign (sw[4] >> 8, w0, 3);
+ break;
+ case 14: sw[3] = amd_bytealign (w0, sw[3] << 16, 2);
+ sw[4] = amd_bytealign (sw[4] >> 16, w0, 2);
+ break;
+ case 15: sw[3] = amd_bytealign (w0, sw[3] << 8, 1);
+ sw[4] = amd_bytealign (sw[4] >> 24, w0, 1);
+ break;
+ case 16: sw[4] = w0;
+ break;
+ case 17: sw[4] = amd_bytealign (w0, sw[4] << 24, 3);
+ sw[5] = amd_bytealign (sw[5] >> 8, w0, 3);
+ break;
+ case 18: sw[4] = amd_bytealign (w0, sw[4] << 16, 2);
+ sw[5] = amd_bytealign (sw[5] >> 16, w0, 2);
+ break;
+ case 19: sw[4] = amd_bytealign (w0, sw[4] << 8, 1);
+ sw[5] = amd_bytealign (sw[5] >> 24, w0, 1);
+ break;
+ case 20: sw[5] = w0;
+ break;
+ case 21: sw[5] = amd_bytealign (w0, sw[5] << 24, 3);
+ sw[6] = amd_bytealign (sw[6] >> 8, w0, 3);
+ break;
+ case 22: sw[5] = amd_bytealign (w0, sw[5] << 16, 2);
+ sw[6] = amd_bytealign (sw[6] >> 16, w0, 2);
+ break;
+ case 23: sw[5] = amd_bytealign (w0, sw[5] << 8, 1);
+ sw[6] = amd_bytealign (sw[6] >> 24, w0, 1);
+ break;
+ case 24: sw[6] = w0;
+ break;
+ case 25: sw[6] = amd_bytealign (w0, sw[6] << 24, 3);
+ sw[7] = amd_bytealign (sw[7] >> 8, w0, 3);
+ break;
+ case 26: sw[6] = amd_bytealign (w0, sw[6] << 16, 2);
+ sw[7] = amd_bytealign (sw[7] >> 16, w0, 2);
+ break;
+ case 27: sw[6] = amd_bytealign (w0, sw[6] << 8, 1);
+ sw[7] = amd_bytealign (sw[7] >> 24, w0, 1);
+ break;
+ case 28: sw[7] = w0;
+ break;
+ case 29: sw[7] = amd_bytealign (w0, sw[7] << 24, 3);
+ sw[8] = amd_bytealign (sw[8] >> 8, w0, 3);
+ break;
+ case 30: sw[7] = amd_bytealign (w0, sw[7] << 16, 2);
+ sw[8] = amd_bytealign (sw[8] >> 16, w0, 2);
+ break;
+ case 31: sw[7] = amd_bytealign (w0, sw[7] << 8, 1);
+ sw[8] = amd_bytealign (sw[8] >> 24, w0, 1);
+ break;
+ }
+ #else
+ switch (salt_len)
+ {
+ case 0: sw[0] = w0;
+ break;
+ case 1: sw[0] = (sw[0] & 0x000000ff) | (w0 << 8);
+ sw[1] = (sw[1] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 2: sw[0] = (sw[0] & 0x0000ffff) | (w0 << 16);
+ sw[1] = (sw[1] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 3: sw[0] = (sw[0] & 0x00ffffff) | (w0 << 24);
+ sw[1] = (sw[1] & 0xff000000) | (w0 >> 8);
+ break;
+ case 4: sw[1] = w0;
+ break;
+ case 5: sw[1] = (sw[1] & 0x000000ff) | (w0 << 8);
+ sw[2] = (sw[2] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 6: sw[1] = (sw[1] & 0x0000ffff) | (w0 << 16);
+ sw[2] = (sw[2] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 7: sw[1] = (sw[1] & 0x00ffffff) | (w0 << 24);
+ sw[2] = (sw[2] & 0xff000000) | (w0 >> 8);
+ break;
+ case 8: sw[2] = w0;
+ break;
+ case 9: sw[2] = (sw[2] & 0x000000ff) | (w0 << 8);
+ sw[3] = (sw[3] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 10: sw[2] = (sw[2] & 0x0000ffff) | (w0 << 16);
+ sw[3] = (sw[3] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 11: sw[2] = (sw[2] & 0x00ffffff) | (w0 << 24);
+ sw[3] = (sw[3] & 0xff000000) | (w0 >> 8);
+ break;
+ case 12: sw[3] = w0;
+ break;
+ case 13: sw[3] = (sw[3] & 0x000000ff) | (w0 << 8);
+ sw[4] = (sw[4] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 14: sw[3] = (sw[3] & 0x0000ffff) | (w0 << 16);
+ sw[4] = (sw[4] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 15: sw[3] = (sw[3] & 0x00ffffff) | (w0 << 24);
+ sw[4] = (sw[4] & 0xff000000) | (w0 >> 8);
+ break;
+ case 16: sw[4] = w0;
+ break;
+ case 17: sw[4] = (sw[4] & 0x000000ff) | (w0 << 8);
+ sw[5] = (sw[5] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 18: sw[4] = (sw[4] & 0x0000ffff) | (w0 << 16);
+ sw[5] = (sw[5] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 19: sw[4] = (sw[4] & 0x00ffffff) | (w0 << 24);
+ sw[5] = (sw[5] & 0xff000000) | (w0 >> 8);
+ break;
+ case 20: sw[5] = w0;
+ break;
+ case 21: sw[5] = (sw[5] & 0x000000ff) | (w0 << 8);
+ sw[6] = (sw[6] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 22: sw[5] = (sw[5] & 0x0000ffff) | (w0 << 16);
+ sw[6] = (sw[6] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 23: sw[5] = (sw[5] & 0x00ffffff) | (w0 << 24);
+ sw[6] = (sw[6] & 0xff000000) | (w0 >> 8);
+ break;
+ case 24: sw[6] = w0;
+ break;
+ case 25: sw[6] = (sw[6] & 0x000000ff) | (w0 << 8);
+ sw[7] = (sw[7] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 26: sw[6] = (sw[6] & 0x0000ffff) | (w0 << 16);
+ sw[7] = (sw[7] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 27: sw[6] = (sw[6] & 0x00ffffff) | (w0 << 24);
+ sw[7] = (sw[7] & 0xff000000) | (w0 >> 8);
+ break;
+ case 28: sw[7] = w0;
+ break;
+ case 29: sw[7] = (sw[7] & 0x000000ff) | (w0 << 8);
+ sw[8] = (sw[8] & 0xffffff00) | (w0 >> 24);
+ break;
+ case 30: sw[7] = (sw[7] & 0x0000ffff) | (w0 << 16);
+ sw[8] = (sw[8] & 0xffff0000) | (w0 >> 16);
+ break;
+ case 31: sw[7] = (sw[7] & 0x00ffffff) | (w0 << 24);
+ sw[8] = (sw[8] & 0xff000000) | (w0 >> 8);
+ break;
+ }
+ #endif
+}
- const int offset_minus_4 = 4 - offset;
+inline void overwrite_at_be (u32x sw[16], const u32x w0, const u32 salt_len)
+{
+ // would be nice to have optimization based on amd_bytealign as with _le counterpart
- switch (offset / 4)
+ switch (salt_len)
{
- case 0:
- w3[2] = amd_bytealign ( 0, w3[1], offset_minus_4);
- w3[1] = amd_bytealign (w3[1], w3[0], offset_minus_4);
- w3[0] = amd_bytealign (w3[0], w2[3], offset_minus_4);
- w2[3] = amd_bytealign (w2[3], w2[2], offset_minus_4);
- w2[2] = amd_bytealign (w2[2], w2[1], offset_minus_4);
- w2[1] = amd_bytealign (w2[1], w2[0], offset_minus_4);
- w2[0] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w1[3] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w1[2] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w1[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w1[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w0[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w0[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w0[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w0[0] = amd_bytealign (w0[0], 0, offset_minus_4);
-
- if (offset_mod_4 == 0)
- {
- w0[0] = w0[1];
- w0[1] = w0[2];
- w0[2] = w0[3];
- w0[3] = w1[0];
- w1[0] = w1[1];
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 1:
- w3[2] = amd_bytealign ( 0, w3[0], offset_minus_4);
- w3[1] = amd_bytealign (w3[0], w2[3], offset_minus_4);
- w3[0] = amd_bytealign (w2[3], w2[2], offset_minus_4);
- w2[3] = amd_bytealign (w2[2], w2[1], offset_minus_4);
- w2[2] = amd_bytealign (w2[1], w2[0], offset_minus_4);
- w2[1] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w2[0] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w1[3] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w1[2] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w1[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w1[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w0[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w0[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w0[1] = amd_bytealign (w0[0], 0, offset_minus_4);
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w0[1] = w0[2];
- w0[2] = w0[3];
- w0[3] = w1[0];
- w1[0] = w1[1];
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 2:
- w3[2] = amd_bytealign ( 0, w2[3], offset_minus_4);
- w3[1] = amd_bytealign (w2[3], w2[2], offset_minus_4);
- w3[0] = amd_bytealign (w2[2], w2[1], offset_minus_4);
- w2[3] = amd_bytealign (w2[1], w2[0], offset_minus_4);
- w2[2] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w2[1] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w2[0] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w1[3] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w1[2] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w1[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w1[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w0[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w0[2] = amd_bytealign (w0[0], 0, offset_minus_4);
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w0[2] = w0[3];
- w0[3] = w1[0];
- w1[0] = w1[1];
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 3:
- w3[2] = amd_bytealign ( 0, w2[2], offset_minus_4);
- w3[1] = amd_bytealign (w2[2], w2[1], offset_minus_4);
- w3[0] = amd_bytealign (w2[1], w2[0], offset_minus_4);
- w2[3] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w2[2] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w2[1] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w2[0] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w1[3] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w1[2] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w1[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w1[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w0[3] = amd_bytealign (w0[0], 0, offset_minus_4);
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w0[3] = w1[0];
- w1[0] = w1[1];
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 4:
- w3[2] = amd_bytealign ( 0, w2[1], offset_minus_4);
- w3[1] = amd_bytealign (w2[1], w2[0], offset_minus_4);
- w3[0] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w2[3] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w2[2] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w2[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w2[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w1[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w1[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w1[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w1[0] = amd_bytealign (w0[0], 0, offset_minus_4);
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w1[0] = w1[1];
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 5:
- w3[2] = amd_bytealign ( 0, w2[0], offset_minus_4);
- w3[1] = amd_bytealign (w2[0], w1[3], offset_minus_4);
- w3[0] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w2[3] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w2[2] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w2[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w2[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w1[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w1[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w1[1] = amd_bytealign (w0[0], 0, offset_minus_4);
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w1[1] = w1[2];
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
+ case 0: sw[0] = w0;
+ break;
+ case 1: sw[0] = (sw[0] & 0xff000000) | (w0 >> 8);
+ sw[1] = (sw[1] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 2: sw[0] = (sw[0] & 0xffff0000) | (w0 >> 16);
+ sw[1] = (sw[1] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 3: sw[0] = (sw[0] & 0xffffff00) | (w0 >> 24);
+ sw[1] = (sw[1] & 0x000000ff) | (w0 << 8);
+ break;
+ case 4: sw[1] = w0;
+ break;
+ case 5: sw[1] = (sw[1] & 0xff000000) | (w0 >> 8);
+ sw[2] = (sw[2] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 6: sw[1] = (sw[1] & 0xffff0000) | (w0 >> 16);
+ sw[2] = (sw[2] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 7: sw[1] = (sw[1] & 0xffffff00) | (w0 >> 24);
+ sw[2] = (sw[2] & 0x000000ff) | (w0 << 8);
+ break;
+ case 8: sw[2] = w0;
+ break;
+ case 9: sw[2] = (sw[2] & 0xff000000) | (w0 >> 8);
+ sw[3] = (sw[3] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 10: sw[2] = (sw[2] & 0xffff0000) | (w0 >> 16);
+ sw[3] = (sw[3] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 11: sw[2] = (sw[2] & 0xffffff00) | (w0 >> 24);
+ sw[3] = (sw[3] & 0x000000ff) | (w0 << 8);
+ break;
+ case 12: sw[3] = w0;
+ break;
+ case 13: sw[3] = (sw[3] & 0xff000000) | (w0 >> 8);
+ sw[4] = (sw[4] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 14: sw[3] = (sw[3] & 0xffff0000) | (w0 >> 16);
+ sw[4] = (sw[4] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 15: sw[3] = (sw[3] & 0xffffff00) | (w0 >> 24);
+ sw[4] = (sw[4] & 0x000000ff) | (w0 << 8);
+ break;
+ case 16: sw[4] = w0;
+ break;
+ case 17: sw[4] = (sw[4] & 0xff000000) | (w0 >> 8);
+ sw[5] = (sw[5] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 18: sw[4] = (sw[4] & 0xffff0000) | (w0 >> 16);
+ sw[5] = (sw[5] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 19: sw[4] = (sw[4] & 0xffffff00) | (w0 >> 24);
+ sw[5] = (sw[5] & 0x000000ff) | (w0 << 8);
+ break;
+ case 20: sw[5] = w0;
+ break;
+ case 21: sw[5] = (sw[5] & 0xff000000) | (w0 >> 8);
+ sw[6] = (sw[6] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 22: sw[5] = (sw[5] & 0xffff0000) | (w0 >> 16);
+ sw[6] = (sw[6] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 23: sw[5] = (sw[5] & 0xffffff00) | (w0 >> 24);
+ sw[6] = (sw[6] & 0x000000ff) | (w0 << 8);
+ break;
+ case 24: sw[6] = w0;
+ break;
+ case 25: sw[6] = (sw[6] & 0xff000000) | (w0 >> 8);
+ sw[7] = (sw[7] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 26: sw[6] = (sw[6] & 0xffff0000) | (w0 >> 16);
+ sw[7] = (sw[7] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 27: sw[6] = (sw[6] & 0xffffff00) | (w0 >> 24);
+ sw[7] = (sw[7] & 0x000000ff) | (w0 << 8);
+ break;
+ case 28: sw[7] = w0;
+ break;
+ case 29: sw[7] = (sw[7] & 0xff000000) | (w0 >> 8);
+ sw[8] = (sw[8] & 0x00ffffff) | (w0 << 24);
+ break;
+ case 30: sw[7] = (sw[7] & 0xffff0000) | (w0 >> 16);
+ sw[8] = (sw[8] & 0x0000ffff) | (w0 << 16);
+ break;
+ case 31: sw[7] = (sw[7] & 0xffffff00) | (w0 >> 24);
+ sw[8] = (sw[8] & 0x000000ff) | (w0 << 8);
+ break;
+ }
+}
- case 6:
- w3[2] = amd_bytealign ( 0, w1[3], offset_minus_4);
- w3[1] = amd_bytealign (w1[3], w1[2], offset_minus_4);
- w3[0] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w2[3] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w2[2] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w2[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w2[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w1[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w1[2] = amd_bytealign (w0[0], 0, offset_minus_4);
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w1[2] = w1[3];
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 7:
- w3[2] = amd_bytealign ( 0, w1[2], offset_minus_4);
- w3[1] = amd_bytealign (w1[2], w1[1], offset_minus_4);
- w3[0] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w2[3] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w2[2] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w2[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w2[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w1[3] = amd_bytealign (w0[0], 0, offset_minus_4);
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w1[3] = w2[0];
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 8:
- w3[2] = amd_bytealign ( 0, w1[1], offset_minus_4);
- w3[1] = amd_bytealign (w1[1], w1[0], offset_minus_4);
- w3[0] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w2[3] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w2[2] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w2[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w2[0] = amd_bytealign (w0[0], 0, offset_minus_4);
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w2[0] = w2[1];
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 9:
- w3[2] = amd_bytealign ( 0, w1[0], offset_minus_4);
- w3[1] = amd_bytealign (w1[0], w0[3], offset_minus_4);
- w3[0] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w2[3] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w2[2] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w2[1] = amd_bytealign (w0[0], 0, offset_minus_4);
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w2[1] = w2[2];
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 10:
- w3[2] = amd_bytealign ( 0, w0[3], offset_minus_4);
- w3[1] = amd_bytealign (w0[3], w0[2], offset_minus_4);
- w3[0] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w2[3] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w2[2] = amd_bytealign (w0[0], 0, offset_minus_4);
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w2[2] = w2[3];
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 11:
- w3[2] = amd_bytealign ( 0, w0[2], offset_minus_4);
- w3[1] = amd_bytealign (w0[2], w0[1], offset_minus_4);
- w3[0] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w2[3] = amd_bytealign (w0[0], 0, offset_minus_4);
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w2[3] = w3[0];
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 12:
- w3[2] = amd_bytealign ( 0, w0[1], offset_minus_4);
- w3[1] = amd_bytealign (w0[1], w0[0], offset_minus_4);
- w3[0] = amd_bytealign (w0[0], 0, offset_minus_4);
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w3[0] = w3[1];
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
-
- case 13:
- w3[2] = amd_bytealign ( 0, w0[0], offset_minus_4);
- w3[1] = amd_bytealign (w0[0], 0, offset_minus_4);
- w3[0] = 0;
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- if (offset_mod_4 == 0)
- {
- w3[1] = w3[2];
- w3[2] = 0;
- }
-
- break;
- }
- #endif
-
- #ifdef IS_NV
- const int offset_minus_4 = 4 - (offset % 4);
-
- const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
-
- switch (offset / 4)
+inline void overwrite_at_le_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x wx, const u32 salt_len)
+{
+ #if defined cl_amd_media_ops
+ switch (salt_len)
{
- case 0:
- w3[1] = __byte_perm (w3[0], w3[1], selector);
- w3[0] = __byte_perm (w2[3], w3[0], selector);
- w2[3] = __byte_perm (w2[2], w2[3], selector);
- w2[2] = __byte_perm (w2[1], w2[2], selector);
- w2[1] = __byte_perm (w2[0], w2[1], selector);
- w2[0] = __byte_perm (w1[3], w2[0], selector);
- w1[3] = __byte_perm (w1[2], w1[3], selector);
- w1[2] = __byte_perm (w1[1], w1[2], selector);
- w1[1] = __byte_perm (w1[0], w1[1], selector);
- w1[0] = __byte_perm (w0[3], w1[0], selector);
- w0[3] = __byte_perm (w0[2], w0[3], selector);
- w0[2] = __byte_perm (w0[1], w0[2], selector);
- w0[1] = __byte_perm (w0[0], w0[1], selector);
- w0[0] = __byte_perm ( 0, w0[0], selector);
-
- break;
-
- case 1:
- w3[1] = __byte_perm (w2[3], w3[0], selector);
- w3[0] = __byte_perm (w2[2], w2[3], selector);
- w2[3] = __byte_perm (w2[1], w2[2], selector);
- w2[2] = __byte_perm (w2[0], w2[1], selector);
- w2[1] = __byte_perm (w1[3], w2[0], selector);
- w2[0] = __byte_perm (w1[2], w1[3], selector);
- w1[3] = __byte_perm (w1[1], w1[2], selector);
- w1[2] = __byte_perm (w1[0], w1[1], selector);
- w1[1] = __byte_perm (w0[3], w1[0], selector);
- w1[0] = __byte_perm (w0[2], w0[3], selector);
- w0[3] = __byte_perm (w0[1], w0[2], selector);
- w0[2] = __byte_perm (w0[0], w0[1], selector);
- w0[1] = __byte_perm ( 0, w0[0], selector);
- w0[0] = 0;
-
- break;
-
- case 2:
- w3[1] = __byte_perm (w2[2], w2[3], selector);
- w3[0] = __byte_perm (w2[1], w2[2], selector);
- w2[3] = __byte_perm (w2[0], w2[1], selector);
- w2[2] = __byte_perm (w1[3], w2[0], selector);
- w2[1] = __byte_perm (w1[2], w1[3], selector);
- w2[0] = __byte_perm (w1[1], w1[2], selector);
- w1[3] = __byte_perm (w1[0], w1[1], selector);
- w1[2] = __byte_perm (w0[3], w1[0], selector);
- w1[1] = __byte_perm (w0[2], w0[3], selector);
- w1[0] = __byte_perm (w0[1], w0[2], selector);
- w0[3] = __byte_perm (w0[0], w0[1], selector);
- w0[2] = __byte_perm ( 0, w0[0], selector);
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 3:
- w3[1] = __byte_perm (w2[1], w2[2], selector);
- w3[0] = __byte_perm (w2[0], w2[1], selector);
- w2[3] = __byte_perm (w1[3], w2[0], selector);
- w2[2] = __byte_perm (w1[2], w1[3], selector);
- w2[1] = __byte_perm (w1[1], w1[2], selector);
- w2[0] = __byte_perm (w1[0], w1[1], selector);
- w1[3] = __byte_perm (w0[3], w1[0], selector);
- w1[2] = __byte_perm (w0[2], w0[3], selector);
- w1[1] = __byte_perm (w0[1], w0[2], selector);
- w1[0] = __byte_perm (w0[0], w0[1], selector);
- w0[3] = __byte_perm ( 0, w0[0], selector);
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 4:
- w3[1] = __byte_perm (w2[0], w2[1], selector);
- w3[0] = __byte_perm (w1[3], w2[0], selector);
- w2[3] = __byte_perm (w1[2], w1[3], selector);
- w2[2] = __byte_perm (w1[1], w1[2], selector);
- w2[1] = __byte_perm (w1[0], w1[1], selector);
- w2[0] = __byte_perm (w0[3], w1[0], selector);
- w1[3] = __byte_perm (w0[2], w0[3], selector);
- w1[2] = __byte_perm (w0[1], w0[2], selector);
- w1[1] = __byte_perm (w0[0], w0[1], selector);
- w1[0] = __byte_perm ( 0, w0[0], selector);
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 5:
- w3[1] = __byte_perm (w1[3], w2[0], selector);
- w3[0] = __byte_perm (w1[2], w1[3], selector);
- w2[3] = __byte_perm (w1[1], w1[2], selector);
- w2[2] = __byte_perm (w1[0], w1[1], selector);
- w2[1] = __byte_perm (w0[3], w1[0], selector);
- w2[0] = __byte_perm (w0[2], w0[3], selector);
- w1[3] = __byte_perm (w0[1], w0[2], selector);
- w1[2] = __byte_perm (w0[0], w0[1], selector);
- w1[1] = __byte_perm ( 0, w0[0], selector);
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 6:
- w3[1] = __byte_perm (w1[2], w1[3], selector);
- w3[0] = __byte_perm (w1[1], w1[2], selector);
- w2[3] = __byte_perm (w1[0], w1[1], selector);
- w2[2] = __byte_perm (w0[3], w1[0], selector);
- w2[1] = __byte_perm (w0[2], w0[3], selector);
- w2[0] = __byte_perm (w0[1], w0[2], selector);
- w1[3] = __byte_perm (w0[0], w0[1], selector);
- w1[2] = __byte_perm ( 0, w0[0], selector);
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 7:
- w3[1] = __byte_perm (w1[1], w1[2], selector);
- w3[0] = __byte_perm (w1[0], w1[1], selector);
- w2[3] = __byte_perm (w0[3], w1[0], selector);
- w2[2] = __byte_perm (w0[2], w0[3], selector);
- w2[1] = __byte_perm (w0[1], w0[2], selector);
- w2[0] = __byte_perm (w0[0], w0[1], selector);
- w1[3] = __byte_perm ( 0, w0[0], selector);
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 8:
- w3[1] = __byte_perm (w1[0], w1[1], selector);
- w3[0] = __byte_perm (w0[3], w1[0], selector);
- w2[3] = __byte_perm (w0[2], w0[3], selector);
- w2[2] = __byte_perm (w0[1], w0[2], selector);
- w2[1] = __byte_perm (w0[0], w0[1], selector);
- w2[0] = __byte_perm ( 0, w0[0], selector);
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 9:
- w3[1] = __byte_perm (w0[3], w1[0], selector);
- w3[0] = __byte_perm (w0[2], w0[3], selector);
- w2[3] = __byte_perm (w0[1], w0[2], selector);
- w2[2] = __byte_perm (w0[0], w0[1], selector);
- w2[1] = __byte_perm ( 0, w0[0], selector);
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 10:
- w3[1] = __byte_perm (w0[2], w0[3], selector);
- w3[0] = __byte_perm (w0[1], w0[2], selector);
- w2[3] = __byte_perm (w0[0], w0[1], selector);
- w2[2] = __byte_perm ( 0, w0[0], selector);
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 11:
- w3[1] = __byte_perm (w0[1], w0[2], selector);
- w3[0] = __byte_perm (w0[0], w0[1], selector);
- w2[3] = __byte_perm ( 0, w0[0], selector);
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 12:
- w3[1] = __byte_perm (w0[0], w0[1], selector);
- w3[0] = __byte_perm ( 0, w0[0], selector);
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
-
- case 13:
- w3[1] = __byte_perm ( 0, w0[0], selector);
- w3[0] = 0;
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
-
- break;
- }
- #endif
-}
-
-static void switch_buffer_by_offset_be (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32 offset)
-{
- #if defined IS_AMD || defined IS_GENERIC
- switch (offset / 4)
- {
- case 0:
- w3[2] = amd_bytealign (w3[1], 0, offset);
- w3[1] = amd_bytealign (w3[0], w3[1], offset);
- w3[0] = amd_bytealign (w2[3], w3[0], offset);
- w2[3] = amd_bytealign (w2[2], w2[3], offset);
- w2[2] = amd_bytealign (w2[1], w2[2], offset);
- w2[1] = amd_bytealign (w2[0], w2[1], offset);
- w2[0] = amd_bytealign (w1[3], w2[0], offset);
- w1[3] = amd_bytealign (w1[2], w1[3], offset);
- w1[2] = amd_bytealign (w1[1], w1[2], offset);
- w1[1] = amd_bytealign (w1[0], w1[1], offset);
- w1[0] = amd_bytealign (w0[3], w1[0], offset);
- w0[3] = amd_bytealign (w0[2], w0[3], offset);
- w0[2] = amd_bytealign (w0[1], w0[2], offset);
- w0[1] = amd_bytealign (w0[0], w0[1], offset);
- w0[0] = amd_bytealign ( 0, w0[0], offset);
- break;
-
- case 1:
- w3[2] = amd_bytealign (w3[0], 0, offset);
- w3[1] = amd_bytealign (w2[3], w3[0], offset);
- w3[0] = amd_bytealign (w2[2], w2[3], offset);
- w2[3] = amd_bytealign (w2[1], w2[2], offset);
- w2[2] = amd_bytealign (w2[0], w2[1], offset);
- w2[1] = amd_bytealign (w1[3], w2[0], offset);
- w2[0] = amd_bytealign (w1[2], w1[3], offset);
- w1[3] = amd_bytealign (w1[1], w1[2], offset);
- w1[2] = amd_bytealign (w1[0], w1[1], offset);
- w1[1] = amd_bytealign (w0[3], w1[0], offset);
- w1[0] = amd_bytealign (w0[2], w0[3], offset);
- w0[3] = amd_bytealign (w0[1], w0[2], offset);
- w0[2] = amd_bytealign (w0[0], w0[1], offset);
- w0[1] = amd_bytealign ( 0, w0[0], offset);
- w0[0] = 0;
- break;
-
- case 2:
- w3[2] = amd_bytealign (w2[3], 0, offset);
- w3[1] = amd_bytealign (w2[2], w2[3], offset);
- w3[0] = amd_bytealign (w2[1], w2[2], offset);
- w2[3] = amd_bytealign (w2[0], w2[1], offset);
- w2[2] = amd_bytealign (w1[3], w2[0], offset);
- w2[1] = amd_bytealign (w1[2], w1[3], offset);
- w2[0] = amd_bytealign (w1[1], w1[2], offset);
- w1[3] = amd_bytealign (w1[0], w1[1], offset);
- w1[2] = amd_bytealign (w0[3], w1[0], offset);
- w1[1] = amd_bytealign (w0[2], w0[3], offset);
- w1[0] = amd_bytealign (w0[1], w0[2], offset);
- w0[3] = amd_bytealign (w0[0], w0[1], offset);
- w0[2] = amd_bytealign ( 0, w0[0], offset);
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 3:
- w3[2] = amd_bytealign (w2[2], 0, offset);
- w3[1] = amd_bytealign (w2[1], w2[2], offset);
- w3[0] = amd_bytealign (w2[0], w2[1], offset);
- w2[3] = amd_bytealign (w1[3], w2[0], offset);
- w2[2] = amd_bytealign (w1[2], w1[3], offset);
- w2[1] = amd_bytealign (w1[1], w1[2], offset);
- w2[0] = amd_bytealign (w1[0], w1[1], offset);
- w1[3] = amd_bytealign (w0[3], w1[0], offset);
- w1[2] = amd_bytealign (w0[2], w0[3], offset);
- w1[1] = amd_bytealign (w0[1], w0[2], offset);
- w1[0] = amd_bytealign (w0[0], w0[1], offset);
- w0[3] = amd_bytealign ( 0, w0[0], offset);
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 4:
- w3[2] = amd_bytealign (w2[1], 0, offset);
- w3[1] = amd_bytealign (w2[0], w2[1], offset);
- w3[0] = amd_bytealign (w1[3], w2[0], offset);
- w2[3] = amd_bytealign (w1[2], w1[3], offset);
- w2[2] = amd_bytealign (w1[1], w1[2], offset);
- w2[1] = amd_bytealign (w1[0], w1[1], offset);
- w2[0] = amd_bytealign (w0[3], w1[0], offset);
- w1[3] = amd_bytealign (w0[2], w0[3], offset);
- w1[2] = amd_bytealign (w0[1], w0[2], offset);
- w1[1] = amd_bytealign (w0[0], w0[1], offset);
- w1[0] = amd_bytealign ( 0, w0[0], offset);
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 5:
- w3[2] = amd_bytealign (w2[0], 0, offset);
- w3[1] = amd_bytealign (w1[3], w2[0], offset);
- w3[0] = amd_bytealign (w1[2], w1[3], offset);
- w2[3] = amd_bytealign (w1[1], w1[2], offset);
- w2[2] = amd_bytealign (w1[0], w1[1], offset);
- w2[1] = amd_bytealign (w0[3], w1[0], offset);
- w2[0] = amd_bytealign (w0[2], w0[3], offset);
- w1[3] = amd_bytealign (w0[1], w0[2], offset);
- w1[2] = amd_bytealign (w0[0], w0[1], offset);
- w1[1] = amd_bytealign ( 0, w0[0], offset);
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 6:
- w3[2] = amd_bytealign (w1[3], 0, offset);
- w3[1] = amd_bytealign (w1[2], w1[3], offset);
- w3[0] = amd_bytealign (w1[1], w1[2], offset);
- w2[3] = amd_bytealign (w1[0], w1[1], offset);
- w2[2] = amd_bytealign (w0[3], w1[0], offset);
- w2[1] = amd_bytealign (w0[2], w0[3], offset);
- w2[0] = amd_bytealign (w0[1], w0[2], offset);
- w1[3] = amd_bytealign (w0[0], w0[1], offset);
- w1[2] = amd_bytealign ( 0, w0[0], offset);
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 7:
- w3[2] = amd_bytealign (w1[2], 0, offset);
- w3[1] = amd_bytealign (w1[1], w1[2], offset);
- w3[0] = amd_bytealign (w1[0], w1[1], offset);
- w2[3] = amd_bytealign (w0[3], w1[0], offset);
- w2[2] = amd_bytealign (w0[2], w0[3], offset);
- w2[1] = amd_bytealign (w0[1], w0[2], offset);
- w2[0] = amd_bytealign (w0[0], w0[1], offset);
- w1[3] = amd_bytealign ( 0, w0[0], offset);
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 8:
- w3[2] = amd_bytealign (w1[1], 0, offset);
- w3[1] = amd_bytealign (w1[0], w1[1], offset);
- w3[0] = amd_bytealign (w0[3], w1[0], offset);
- w2[3] = amd_bytealign (w0[2], w0[3], offset);
- w2[2] = amd_bytealign (w0[1], w0[2], offset);
- w2[1] = amd_bytealign (w0[0], w0[1], offset);
- w2[0] = amd_bytealign ( 0, w0[0], offset);
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 9:
- w3[2] = amd_bytealign (w1[0], 0, offset);
- w3[1] = amd_bytealign (w0[3], w1[0], offset);
- w3[0] = amd_bytealign (w0[2], w0[3], offset);
- w2[3] = amd_bytealign (w0[1], w0[2], offset);
- w2[2] = amd_bytealign (w0[0], w0[1], offset);
- w2[1] = amd_bytealign ( 0, w0[0], offset);
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 10:
- w3[2] = amd_bytealign (w0[3], 0, offset);
- w3[1] = amd_bytealign (w0[2], w0[3], offset);
- w3[0] = amd_bytealign (w0[1], w0[2], offset);
- w2[3] = amd_bytealign (w0[0], w0[1], offset);
- w2[2] = amd_bytealign ( 0, w0[0], offset);
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 11:
- w3[2] = amd_bytealign (w0[2], 0, offset);
- w3[1] = amd_bytealign (w0[1], w0[2], offset);
- w3[0] = amd_bytealign (w0[0], w0[1], offset);
- w2[3] = amd_bytealign ( 0, w0[0], offset);
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 12:
- w3[2] = amd_bytealign (w0[1], 0, offset);
- w3[1] = amd_bytealign (w0[0], w0[1], offset);
- w3[0] = amd_bytealign ( 0, w0[0], offset);
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 13:
- w3[2] = amd_bytealign (w0[0], 0, offset);
- w3[1] = amd_bytealign ( 0, w0[0], offset);
- w3[0] = 0;
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
- }
- #endif
-
- #ifdef IS_NV
- const int selector = (0x76543210 >> ((offset & 3) * 4)) & 0xffff;
-
- switch (offset / 4)
- {
- case 0:
- w3[1] = __byte_perm (w3[1], w3[0], selector);
- w3[0] = __byte_perm (w3[0], w2[3], selector);
- w2[3] = __byte_perm (w2[3], w2[2], selector);
- w2[2] = __byte_perm (w2[2], w2[1], selector);
- w2[1] = __byte_perm (w2[1], w2[0], selector);
- w2[0] = __byte_perm (w2[0], w1[3], selector);
- w1[3] = __byte_perm (w1[3], w1[2], selector);
- w1[2] = __byte_perm (w1[2], w1[1], selector);
- w1[1] = __byte_perm (w1[1], w1[0], selector);
- w1[0] = __byte_perm (w1[0], w0[3], selector);
- w0[3] = __byte_perm (w0[3], w0[2], selector);
- w0[2] = __byte_perm (w0[2], w0[1], selector);
- w0[1] = __byte_perm (w0[1], w0[0], selector);
- w0[0] = __byte_perm (w0[0], 0, selector);
- break;
-
- case 1:
- w3[1] = __byte_perm (w3[0], w2[3], selector);
- w3[0] = __byte_perm (w2[3], w2[2], selector);
- w2[3] = __byte_perm (w2[2], w2[1], selector);
- w2[2] = __byte_perm (w2[1], w2[0], selector);
- w2[1] = __byte_perm (w2[0], w1[3], selector);
- w2[0] = __byte_perm (w1[3], w1[2], selector);
- w1[3] = __byte_perm (w1[2], w1[1], selector);
- w1[2] = __byte_perm (w1[1], w1[0], selector);
- w1[1] = __byte_perm (w1[0], w0[3], selector);
- w1[0] = __byte_perm (w0[3], w0[2], selector);
- w0[3] = __byte_perm (w0[2], w0[1], selector);
- w0[2] = __byte_perm (w0[1], w0[0], selector);
- w0[1] = __byte_perm (w0[0], 0, selector);
- w0[0] = 0;
- break;
-
- case 2:
- w3[1] = __byte_perm (w2[3], w2[2], selector);
- w3[0] = __byte_perm (w2[2], w2[1], selector);
- w2[3] = __byte_perm (w2[1], w2[0], selector);
- w2[2] = __byte_perm (w2[0], w1[3], selector);
- w2[1] = __byte_perm (w1[3], w1[2], selector);
- w2[0] = __byte_perm (w1[2], w1[1], selector);
- w1[3] = __byte_perm (w1[1], w1[0], selector);
- w1[2] = __byte_perm (w1[0], w0[3], selector);
- w1[1] = __byte_perm (w0[3], w0[2], selector);
- w1[0] = __byte_perm (w0[2], w0[1], selector);
- w0[3] = __byte_perm (w0[1], w0[0], selector);
- w0[2] = __byte_perm (w0[0], 0, selector);
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 3:
- w3[1] = __byte_perm (w2[2], w2[1], selector);
- w3[0] = __byte_perm (w2[1], w2[0], selector);
- w2[3] = __byte_perm (w2[0], w1[3], selector);
- w2[2] = __byte_perm (w1[3], w1[2], selector);
- w2[1] = __byte_perm (w1[2], w1[1], selector);
- w2[0] = __byte_perm (w1[1], w1[0], selector);
- w1[3] = __byte_perm (w1[0], w0[3], selector);
- w1[2] = __byte_perm (w0[3], w0[2], selector);
- w1[1] = __byte_perm (w0[2], w0[1], selector);
- w1[0] = __byte_perm (w0[1], w0[0], selector);
- w0[3] = __byte_perm (w0[0], 0, selector);
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 4:
- w3[1] = __byte_perm (w2[1], w2[0], selector);
- w3[0] = __byte_perm (w2[0], w1[3], selector);
- w2[3] = __byte_perm (w1[3], w1[2], selector);
- w2[2] = __byte_perm (w1[2], w1[1], selector);
- w2[1] = __byte_perm (w1[1], w1[0], selector);
- w2[0] = __byte_perm (w1[0], w0[3], selector);
- w1[3] = __byte_perm (w0[3], w0[2], selector);
- w1[2] = __byte_perm (w0[2], w0[1], selector);
- w1[1] = __byte_perm (w0[1], w0[0], selector);
- w1[0] = __byte_perm (w0[0], 0, selector);
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 5:
- w3[1] = __byte_perm (w2[0], w1[3], selector);
- w3[0] = __byte_perm (w1[3], w1[2], selector);
- w2[3] = __byte_perm (w1[2], w1[1], selector);
- w2[2] = __byte_perm (w1[1], w1[0], selector);
- w2[1] = __byte_perm (w1[0], w0[3], selector);
- w2[0] = __byte_perm (w0[3], w0[2], selector);
- w1[3] = __byte_perm (w0[2], w0[1], selector);
- w1[2] = __byte_perm (w0[1], w0[0], selector);
- w1[1] = __byte_perm (w0[0], 0, selector);
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 6:
- w3[1] = __byte_perm (w1[3], w1[2], selector);
- w3[0] = __byte_perm (w1[2], w1[1], selector);
- w2[3] = __byte_perm (w1[1], w1[0], selector);
- w2[2] = __byte_perm (w1[0], w0[3], selector);
- w2[1] = __byte_perm (w0[3], w0[2], selector);
- w2[0] = __byte_perm (w0[2], w0[1], selector);
- w1[3] = __byte_perm (w0[1], w0[0], selector);
- w1[2] = __byte_perm (w0[0], 0, selector);
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 7:
- w3[1] = __byte_perm (w1[2], w1[1], selector);
- w3[0] = __byte_perm (w1[1], w1[0], selector);
- w2[3] = __byte_perm (w1[0], w0[3], selector);
- w2[2] = __byte_perm (w0[3], w0[2], selector);
- w2[1] = __byte_perm (w0[2], w0[1], selector);
- w2[0] = __byte_perm (w0[1], w0[0], selector);
- w1[3] = __byte_perm (w0[0], 0, selector);
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 8:
- w3[1] = __byte_perm (w1[1], w1[0], selector);
- w3[0] = __byte_perm (w1[0], w0[3], selector);
- w2[3] = __byte_perm (w0[3], w0[2], selector);
- w2[2] = __byte_perm (w0[2], w0[1], selector);
- w2[1] = __byte_perm (w0[1], w0[0], selector);
- w2[0] = __byte_perm (w0[0], 0, selector);
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 9:
- w3[1] = __byte_perm (w1[0], w0[3], selector);
- w3[0] = __byte_perm (w0[3], w0[2], selector);
- w2[3] = __byte_perm (w0[2], w0[1], selector);
- w2[2] = __byte_perm (w0[1], w0[0], selector);
- w2[1] = __byte_perm (w0[0], 0, selector);
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 10:
- w3[1] = __byte_perm (w0[3], w0[2], selector);
- w3[0] = __byte_perm (w0[2], w0[1], selector);
- w2[3] = __byte_perm (w0[1], w0[0], selector);
- w2[2] = __byte_perm (w0[0], 0, selector);
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 11:
- w3[1] = __byte_perm (w0[2], w0[1], selector);
- w3[0] = __byte_perm (w0[1], w0[0], selector);
- w2[3] = __byte_perm (w0[0], 0, selector);
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 12:
- w3[1] = __byte_perm (w0[1], w0[0], selector);
- w3[0] = __byte_perm (w0[0], 0, selector);
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
-
- case 13:
- w3[1] = __byte_perm (w0[0], 0, selector);
- w3[0] = 0;
- w2[3] = 0;
- w2[2] = 0;
- w2[1] = 0;
- w2[0] = 0;
- w1[3] = 0;
- w1[2] = 0;
- w1[1] = 0;
- w1[0] = 0;
- w0[3] = 0;
- w0[2] = 0;
- w0[1] = 0;
- w0[0] = 0;
- break;
+ case 0: w0[0] = wx;
+ break;
+ case 1: w0[0] = amd_bytealign (wx, w0[0] << 24, 3);
+ w0[1] = amd_bytealign (w0[1] >> 8, wx, 3);
+ break;
+ case 2: w0[0] = amd_bytealign (wx, w0[0] << 16, 2);
+ w0[1] = amd_bytealign (w0[1] >> 16, wx, 2);
+ break;
+ case 3: w0[0] = amd_bytealign (wx, w0[0] << 8, 1);
+ w0[1] = amd_bytealign (w0[1] >> 24, wx, 1);
+ break;
+ case 4: w0[1] = wx;
+ break;
+ case 5: w0[1] = amd_bytealign (wx, w0[1] << 24, 3);
+ w0[2] = amd_bytealign (w0[2] >> 8, wx, 3);
+ break;
+ case 6: w0[1] = amd_bytealign (wx, w0[1] << 16, 2);
+ w0[2] = amd_bytealign (w0[2] >> 16, wx, 2);
+ break;
+ case 7: w0[1] = amd_bytealign (wx, w0[1] << 8, 1);
+ w0[2] = amd_bytealign (w0[2] >> 24, wx, 1);
+ break;
+ case 8: w0[2] = wx;
+ break;
+ case 9: w0[2] = amd_bytealign (wx, w0[2] << 24, 3);
+ w0[3] = amd_bytealign (w0[3] >> 8, wx, 3);
+ break;
+ case 10: w0[2] = amd_bytealign (wx, w0[2] << 16, 2);
+ w0[3] = amd_bytealign (w0[3] >> 16, wx, 2);
+ break;
+ case 11: w0[2] = amd_bytealign (wx, w0[2] << 8, 1);
+ w0[3] = amd_bytealign (w0[3] >> 24, wx, 1);
+ break;
+ case 12: w0[3] = wx;
+ break;
+ case 13: w0[3] = amd_bytealign (wx, w0[3] << 24, 3);
+ w1[0] = amd_bytealign (w1[0] >> 8, wx, 3);
+ break;
+ case 14: w0[3] = amd_bytealign (wx, w0[3] << 16, 2);
+ w1[0] = amd_bytealign (w1[0] >> 16, wx, 2);
+ break;
+ case 15: w0[3] = amd_bytealign (wx, w0[3] << 8, 1);
+ w1[0] = amd_bytealign (w1[0] >> 24, wx, 1);
+ break;
+ case 16: w1[0] = wx;
+ break;
+ case 17: w1[0] = amd_bytealign (wx, w1[0] << 24, 3);
+ w1[1] = amd_bytealign (w1[1] >> 8, wx, 3);
+ break;
+ case 18: w1[0] = amd_bytealign (wx, w1[0] << 16, 2);
+ w1[1] = amd_bytealign (w1[1] >> 16, wx, 2);
+ break;
+ case 19: w1[0] = amd_bytealign (wx, w1[0] << 8, 1);
+ w1[1] = amd_bytealign (w1[1] >> 24, wx, 1);
+ break;
+ case 20: w1[1] = wx;
+ break;
+ case 21: w1[1] = amd_bytealign (wx, w1[1] << 24, 3);
+ w1[2] = amd_bytealign (w1[2] >> 8, wx, 3);
+ break;
+ case 22: w1[1] = amd_bytealign (wx, w1[1] << 16, 2);
+ w1[2] = amd_bytealign (w1[2] >> 16, wx, 2);
+ break;
+ case 23: w1[1] = amd_bytealign (wx, w1[1] << 8, 1);
+ w1[2] = amd_bytealign (w1[2] >> 24, wx, 1);
+ break;
+ case 24: w1[2] = wx;
+ break;
+ case 25: w1[2] = amd_bytealign (wx, w1[2] << 24, 3);
+ w1[3] = amd_bytealign (w1[3] >> 8, wx, 3);
+ break;
+ case 26: w1[2] = amd_bytealign (wx, w1[2] << 16, 2);
+ w1[3] = amd_bytealign (w1[3] >> 16, wx, 2);
+ break;
+ case 27: w1[2] = amd_bytealign (wx, w1[2] << 8, 1);
+ w1[3] = amd_bytealign (w1[3] >> 24, wx, 1);
+ break;
+ case 28: w1[3] = wx;
+ break;
+ case 29: w1[3] = amd_bytealign (wx, w1[3] << 24, 3);
+ w2[0] = amd_bytealign (w2[0] >> 8, wx, 3);
+ break;
+ case 30: w1[3] = amd_bytealign (wx, w1[3] << 16, 2);
+ w2[0] = amd_bytealign (w2[0] >> 16, wx, 2);
+ break;
+ case 31: w1[3] = amd_bytealign (wx, w1[3] << 8, 1);
+ w2[0] = amd_bytealign (w2[0] >> 24, wx, 1);
+ break;
+ case 32: w2[0] = wx;
+ break;
+ case 33: w2[0] = amd_bytealign (wx, w2[0] << 24, 3);
+ w2[1] = amd_bytealign (w2[1] >> 8, wx, 3);
+ break;
+ case 34: w2[0] = amd_bytealign (wx, w2[0] << 16, 2);
+ w2[1] = amd_bytealign (w2[1] >> 16, wx, 2);
+ break;
+ case 35: w2[0] = amd_bytealign (wx, w2[0] << 8, 1);
+ w2[1] = amd_bytealign (w2[1] >> 24, wx, 1);
+ break;
+ case 36: w2[1] = wx;
+ break;
+ case 37: w2[1] = amd_bytealign (wx, w2[1] << 24, 3);
+ w2[2] = amd_bytealign (w2[2] >> 8, wx, 3);
+ break;
+ case 38: w2[1] = amd_bytealign (wx, w2[1] << 16, 2);
+ w2[2] = amd_bytealign (w2[2] >> 16, wx, 2);
+ break;
+ case 39: w2[1] = amd_bytealign (wx, w2[1] << 8, 1);
+ w2[2] = amd_bytealign (w2[2] >> 24, wx, 1);
+ break;
+ case 40: w2[2] = wx;
+ break;
+ case 41: w2[2] = amd_bytealign (wx, w2[2] << 24, 3);
+ w2[3] = amd_bytealign (w2[3] >> 8, wx, 3);
+ break;
+ case 42: w2[2] = amd_bytealign (wx, w2[2] << 16, 2);
+ w2[3] = amd_bytealign (w2[3] >> 16, wx, 2);
+ break;
+ case 43: w2[2] = amd_bytealign (wx, w2[2] << 8, 1);
+ w2[3] = amd_bytealign (w2[3] >> 24, wx, 1);
+ break;
+ case 44: w2[3] = wx;
+ break;
+ case 45: w2[3] = amd_bytealign (wx, w2[3] << 24, 3);
+ w3[0] = amd_bytealign (w3[0] >> 8, wx, 3);
+ break;
+ case 46: w2[3] = amd_bytealign (wx, w2[3] << 16, 2);
+ w3[0] = amd_bytealign (w3[0] >> 16, wx, 2);
+ break;
+ case 47: w2[3] = amd_bytealign (wx, w2[3] << 8, 1);
+ w3[0] = amd_bytealign (w3[0] >> 24, wx, 1);
+ break;
+ case 48: w3[0] = wx;
+ break;
+ case 49: w3[0] = amd_bytealign (wx, w3[0] << 24, 3);
+ w3[1] = amd_bytealign (w3[1] >> 8, wx, 3);
+ break;
+ case 50: w3[0] = amd_bytealign (wx, w3[0] << 16, 2);
+ w3[1] = amd_bytealign (w3[1] >> 16, wx, 2);
+ break;
+ case 51: w3[0] = amd_bytealign (wx, w3[0] << 8, 1);
+ w3[1] = amd_bytealign (w3[1] >> 24, wx, 1);
+ break;
+ case 52: w3[1] = wx;
+ break;
+ case 53: w3[1] = amd_bytealign (wx, w3[1] << 24, 3);
+ w3[2] = amd_bytealign (w3[2] >> 8, wx, 3);
+ break;
+ case 54: w3[1] = amd_bytealign (wx, w3[1] << 16, 2);
+ w3[2] = amd_bytealign (w3[2] >> 16, wx, 2);
+ break;
+ case 55: w3[1] = amd_bytealign (wx, w3[1] << 8, 1);
+ w3[2] = amd_bytealign (w3[2] >> 24, wx, 1);
+ break;
+ case 56: w3[2] = wx;
+ break;
+ case 57: w3[2] = amd_bytealign (wx, w3[2] << 24, 3);
+ w3[3] = amd_bytealign (w3[3] >> 8, wx, 3);
+ break;
+ case 58: w3[2] = amd_bytealign (wx, w3[2] << 16, 2);
+ w3[3] = amd_bytealign (w3[3] >> 16, wx, 2);
+ break;
+ case 59: w3[2] = amd_bytealign (wx, w3[2] << 8, 1);
+ w3[3] = amd_bytealign (w3[3] >> 24, wx, 1);
+ break;
+ case 60: w3[3] = wx;
+ break;
+ case 61: w3[3] = amd_bytealign (wx, w3[3] << 24, 3);
+ //w4[0] = amd_bytealign (w4[0] >> 8, wx, 3);
+ break;
+ case 62: w3[3] = amd_bytealign (wx, w3[3] << 16, 2);
+ //w4[0] = amd_bytealign (w4[0] >> 16, wx, 2);
+ break;
+ case 63: w3[3] = amd_bytealign (wx, w3[3] << 8, 1);
+ //w4[0] = amd_bytealign (w4[0] >> 24, wx, 1);
+ break;
+ }
+ #else
+ switch (salt_len)
+ {
+ case 0: w0[0] = wx;
+ break;
+ case 1: w0[0] = (w0[0] & 0x000000ff) | (wx << 8);
+ w0[1] = (w0[1] & 0xffffff00) | (wx >> 24);
+ break;
+ case 2: w0[0] = (w0[0] & 0x0000ffff) | (wx << 16);
+ w0[1] = (w0[1] & 0xffff0000) | (wx >> 16);
+ break;
+ case 3: w0[0] = (w0[0] & 0x00ffffff) | (wx << 24);
+ w0[1] = (w0[1] & 0xff000000) | (wx >> 8);
+ break;
+ case 4: w0[1] = wx;
+ break;
+ case 5: w0[1] = (w0[1] & 0x000000ff) | (wx << 8);
+ w0[2] = (w0[2] & 0xffffff00) | (wx >> 24);
+ break;
+ case 6: w0[1] = (w0[1] & 0x0000ffff) | (wx << 16);
+ w0[2] = (w0[2] & 0xffff0000) | (wx >> 16);
+ break;
+ case 7: w0[1] = (w0[1] & 0x00ffffff) | (wx << 24);
+ w0[2] = (w0[2] & 0xff000000) | (wx >> 8);
+ break;
+ case 8: w0[2] = wx;
+ break;
+ case 9: w0[2] = (w0[2] & 0x000000ff) | (wx << 8);
+ w0[3] = (w0[3] & 0xffffff00) | (wx >> 24);
+ break;
+ case 10: w0[2] = (w0[2] & 0x0000ffff) | (wx << 16);
+ w0[3] = (w0[3] & 0xffff0000) | (wx >> 16);
+ break;
+ case 11: w0[2] = (w0[2] & 0x00ffffff) | (wx << 24);
+ w0[3] = (w0[3] & 0xff000000) | (wx >> 8);
+ break;
+ case 12: w0[3] = wx;
+ break;
+ case 13: w0[3] = (w0[3] & 0x000000ff) | (wx << 8);
+ w1[0] = (w1[0] & 0xffffff00) | (wx >> 24);
+ break;
+ case 14: w0[3] = (w0[3] & 0x0000ffff) | (wx << 16);
+ w1[0] = (w1[0] & 0xffff0000) | (wx >> 16);
+ break;
+ case 15: w0[3] = (w0[3] & 0x00ffffff) | (wx << 24);
+ w1[0] = (w1[0] & 0xff000000) | (wx >> 8);
+ break;
+ case 16: w1[0] = wx;
+ break;
+ case 17: w1[0] = (w1[0] & 0x000000ff) | (wx << 8);
+ w1[1] = (w1[1] & 0xffffff00) | (wx >> 24);
+ break;
+ case 18: w1[0] = (w1[0] & 0x0000ffff) | (wx << 16);
+ w1[1] = (w1[1] & 0xffff0000) | (wx >> 16);
+ break;
+ case 19: w1[0] = (w1[0] & 0x00ffffff) | (wx << 24);
+ w1[1] = (w1[1] & 0xff000000) | (wx >> 8);
+ break;
+ case 20: w1[1] = wx;
+ break;
+ case 21: w1[1] = (w1[1] & 0x000000ff) | (wx << 8);
+ w1[2] = (w1[2] & 0xffffff00) | (wx >> 24);
+ break;
+ case 22: w1[1] = (w1[1] & 0x0000ffff) | (wx << 16);
+ w1[2] = (w1[2] & 0xffff0000) | (wx >> 16);
+ break;
+ case 23: w1[1] = (w1[1] & 0x00ffffff) | (wx << 24);
+ w1[2] = (w1[2] & 0xff000000) | (wx >> 8);
+ break;
+ case 24: w1[2] = wx;
+ break;
+ case 25: w1[2] = (w1[2] & 0x000000ff) | (wx << 8);
+ w1[3] = (w1[3] & 0xffffff00) | (wx >> 24);
+ break;
+ case 26: w1[2] = (w1[2] & 0x0000ffff) | (wx << 16);
+ w1[3] = (w1[3] & 0xffff0000) | (wx >> 16);
+ break;
+ case 27: w1[2] = (w1[2] & 0x00ffffff) | (wx << 24);
+ w1[3] = (w1[3] & 0xff000000) | (wx >> 8);
+ break;
+ case 28: w1[3] = wx;
+ break;
+ case 29: w1[3] = (w1[3] & 0x000000ff) | (wx << 8);
+ w2[0] = (w2[0] & 0xffffff00) | (wx >> 24);
+ break;
+ case 30: w1[3] = (w1[3] & 0x0000ffff) | (wx << 16);
+ w2[0] = (w2[0] & 0xffff0000) | (wx >> 16);
+ break;
+ case 31: w1[3] = (w1[3] & 0x00ffffff) | (wx << 24);
+ w2[0] = (w2[0] & 0xff000000) | (wx >> 8);
+ break;
+ case 32: w2[0] = wx;
+ break;
+ case 33: w2[0] = (w2[0] & 0x000000ff) | (wx << 8);
+ w2[1] = (w2[1] & 0xffffff00) | (wx >> 24);
+ break;
+ case 34: w2[0] = (w2[0] & 0x0000ffff) | (wx << 16);
+ w2[1] = (w2[1] & 0xffff0000) | (wx >> 16);
+ break;
+ case 35: w2[0] = (w2[0] & 0x00ffffff) | (wx << 24);
+ w2[1] = (w2[1] & 0xff000000) | (wx >> 8);
+ break;
+ case 36: w2[1] = wx;
+ break;
+ case 37: w2[1] = (w2[1] & 0x000000ff) | (wx << 8);
+ w2[2] = (w2[2] & 0xffffff00) | (wx >> 24);
+ break;
+ case 38: w2[1] = (w2[1] & 0x0000ffff) | (wx << 16);
+ w2[2] = (w2[2] & 0xffff0000) | (wx >> 16);
+ break;
+ case 39: w2[1] = (w2[1] & 0x00ffffff) | (wx << 24);
+ w2[2] = (w2[2] & 0xff000000) | (wx >> 8);
+ break;
+ case 40: w2[2] = wx;
+ break;
+ case 41: w2[2] = (w2[2] & 0x000000ff) | (wx << 8);
+ w2[3] = (w2[3] & 0xffffff00) | (wx >> 24);
+ break;
+ case 42: w2[2] = (w2[2] & 0x0000ffff) | (wx << 16);
+ w2[3] = (w2[3] & 0xffff0000) | (wx >> 16);
+ break;
+ case 43: w2[2] = (w2[2] & 0x00ffffff) | (wx << 24);
+ w2[3] = (w2[3] & 0xff000000) | (wx >> 8);
+ break;
+ case 44: w2[3] = wx;
+ break;
+ case 45: w2[3] = (w2[3] & 0x000000ff) | (wx << 8);
+ w3[0] = (w3[0] & 0xffffff00) | (wx >> 24);
+ break;
+ case 46: w2[3] = (w2[3] & 0x0000ffff) | (wx << 16);
+ w3[0] = (w3[0] & 0xffff0000) | (wx >> 16);
+ break;
+ case 47: w2[3] = (w2[3] & 0x00ffffff) | (wx << 24);
+ w3[0] = (w3[0] & 0xff000000) | (wx >> 8);
+ break;
+ case 48: w3[0] = wx;
+ break;
+ case 49: w3[0] = (w3[0] & 0x000000ff) | (wx << 8);
+ w3[1] = (w3[1] & 0xffffff00) | (wx >> 24);
+ break;
+ case 50: w3[0] = (w3[0] & 0x0000ffff) | (wx << 16);
+ w3[1] = (w3[1] & 0xffff0000) | (wx >> 16);
+ break;
+ case 51: w3[0] = (w3[0] & 0x00ffffff) | (wx << 24);
+ w3[1] = (w3[1] & 0xff000000) | (wx >> 8);
+ break;
+ case 52: w3[1] = wx;
+ break;
+ case 53: w3[1] = (w3[1] & 0x000000ff) | (wx << 8);
+ w3[2] = (w3[2] & 0xffffff00) | (wx >> 24);
+ break;
+ case 54: w3[1] = (w3[1] & 0x0000ffff) | (wx << 16);
+ w3[2] = (w3[2] & 0xffff0000) | (wx >> 16);
+ break;
+ case 55: w3[1] = (w3[1] & 0x00ffffff) | (wx << 24);
+ w3[2] = (w3[2] & 0xff000000) | (wx >> 8);
+ break;
+ case 56: w3[2] = wx;
+ break;
+ case 57: w3[2] = (w3[2] & 0x000000ff) | (wx << 8);
+ w3[3] = (w3[3] & 0xffffff00) | (wx >> 24);
+ break;
+ case 58: w3[2] = (w3[2] & 0x0000ffff) | (wx << 16);
+ w3[3] = (w3[3] & 0xffff0000) | (wx >> 16);
+ break;
+ case 59: w3[2] = (w3[2] & 0x00ffffff) | (wx << 24);
+ w3[3] = (w3[3] & 0xff000000) | (wx >> 8);
+ break;
+ case 60: w3[3] = wx;
+ break;
+ case 61: w3[3] = (w3[3] & 0x000000ff) | (wx << 8);
+ //w4[0] = (w4[0] & 0xffffff00) | (wx >> 24);
+ break;
+ case 62: w3[3] = (w3[3] & 0x0000ffff) | (wx << 16);
+ //w4[0] = (w4[0] & 0xffff0000) | (wx >> 16);
+ break;
+ case 63: w3[3] = (w3[3] & 0x00ffffff) | (wx << 24);
+ //w4[0] = (w4[0] & 0xff000000) | (wx >> 8);
+ break;
}
#endif
}
-/* not needed anymore?
-
-// before: append_0x80_2_be
-static void append_0x80_2x4_be (u32 w0[4], u32 w1[4], const u32 offset)
+inline void overwrite_at_be_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x wx, const u32 salt_len)
{
- switch (offset)
- {
- case 0:
- w0[0] |= 0x80000000;
- break;
-
- case 1:
- w0[0] |= 0x800000;
- break;
-
- case 2:
- w0[0] |= 0x8000;
- break;
-
- case 3:
- w0[0] |= 0x80;
- break;
-
- case 4:
- w0[1] |= 0x80000000;
- break;
-
- case 5:
- w0[1] |= 0x800000;
- break;
-
- case 6:
- w0[1] |= 0x8000;
- break;
-
- case 7:
- w0[1] |= 0x80;
- break;
+ // would be nice to have optimization based on amd_bytealign as with _le counterpart
- case 8:
- w0[2] |= 0x80000000;
- break;
-
- case 9:
- w0[2] |= 0x800000;
- break;
-
- case 10:
- w0[2] |= 0x8000;
- break;
-
- case 11:
- w0[2] |= 0x80;
- break;
-
- case 12:
- w0[3] |= 0x80000000;
- break;
-
- case 13:
- w0[3] |= 0x800000;
- break;
-
- case 14:
- w0[3] |= 0x8000;
- break;
-
- case 15:
- w0[3] |= 0x80;
- break;
-
- case 16:
- w1[0] |= 0x80000000;
- break;
-
- case 17:
- w1[0] |= 0x800000;
- break;
-
- case 18:
- w1[0] |= 0x8000;
- break;
-
- case 19:
- w1[0] |= 0x80;
- break;
-
- case 20:
- w1[1] |= 0x80000000;
- break;
-
- case 21:
- w1[1] |= 0x800000;
- break;
-
- case 22:
- w1[1] |= 0x8000;
- break;
-
- case 23:
- w1[1] |= 0x80;
- break;
-
- case 24:
- w1[2] |= 0x80000000;
- break;
-
- case 25:
- w1[2] |= 0x800000;
- break;
-
- case 26:
- w1[2] |= 0x8000;
- break;
-
- case 27:
- w1[2] |= 0x80;
- break;
-
- case 28:
- w1[3] |= 0x80000000;
- break;
-
- case 29:
- w1[3] |= 0x800000;
- break;
-
- case 30:
- w1[3] |= 0x8000;
- break;
-
- case 31:
- w1[3] |= 0x80;
- break;
+ switch (salt_len)
+ {
+ case 0: w0[0] = wx;
+ break;
+ case 1: w0[0] = (w0[0] & 0xff000000) | (wx >> 8);
+ w0[1] = (w0[1] & 0x00ffffff) | (wx << 24);
+ break;
+ case 2: w0[0] = (w0[0] & 0xffff0000) | (wx >> 16);
+ w0[1] = (w0[1] & 0x0000ffff) | (wx << 16);
+ break;
+ case 3: w0[0] = (w0[0] & 0xffffff00) | (wx >> 24);
+ w0[1] = (w0[1] & 0x000000ff) | (wx << 8);
+ break;
+ case 4: w0[1] = wx;
+ break;
+ case 5: w0[1] = (w0[1] & 0xff000000) | (wx >> 8);
+ w0[2] = (w0[2] & 0x00ffffff) | (wx << 24);
+ break;
+ case 6: w0[1] = (w0[1] & 0xffff0000) | (wx >> 16);
+ w0[2] = (w0[2] & 0x0000ffff) | (wx << 16);
+ break;
+ case 7: w0[1] = (w0[1] & 0xffffff00) | (wx >> 24);
+ w0[2] = (w0[2] & 0x000000ff) | (wx << 8);
+ break;
+ case 8: w0[2] = wx;
+ break;
+ case 9: w0[2] = (w0[2] & 0xff000000) | (wx >> 8);
+ w0[3] = (w0[3] & 0x00ffffff) | (wx << 24);
+ break;
+ case 10: w0[2] = (w0[2] & 0xffff0000) | (wx >> 16);
+ w0[3] = (w0[3] & 0x0000ffff) | (wx << 16);
+ break;
+ case 11: w0[2] = (w0[2] & 0xffffff00) | (wx >> 24);
+ w0[3] = (w0[3] & 0x000000ff) | (wx << 8);
+ break;
+ case 12: w0[3] = wx;
+ break;
+ case 13: w0[3] = (w0[3] & 0xff000000) | (wx >> 8);
+ w1[0] = (w1[0] & 0x00ffffff) | (wx << 24);
+ break;
+ case 14: w0[3] = (w0[3] & 0xffff0000) | (wx >> 16);
+ w1[0] = (w1[0] & 0x0000ffff) | (wx << 16);
+ break;
+ case 15: w0[3] = (w0[3] & 0xffffff00) | (wx >> 24);
+ w1[0] = (w1[0] & 0x000000ff) | (wx << 8);
+ break;
+ case 16: w1[0] = wx;
+ break;
+ case 17: w1[0] = (w1[0] & 0xff000000) | (wx >> 8);
+ w1[1] = (w1[1] & 0x00ffffff) | (wx << 24);
+ break;
+ case 18: w1[0] = (w1[0] & 0xffff0000) | (wx >> 16);
+ w1[1] = (w1[1] & 0x0000ffff) | (wx << 16);
+ break;
+ case 19: w1[0] = (w1[0] & 0xffffff00) | (wx >> 24);
+ w1[1] = (w1[1] & 0x000000ff) | (wx << 8);
+ break;
+ case 20: w1[1] = wx;
+ break;
+ case 21: w1[1] = (w1[1] & 0xff000000) | (wx >> 8);
+ w1[2] = (w1[2] & 0x00ffffff) | (wx << 24);
+ break;
+ case 22: w1[1] = (w1[1] & 0xffff0000) | (wx >> 16);
+ w1[2] = (w1[2] & 0x0000ffff) | (wx << 16);
+ break;
+ case 23: w1[1] = (w1[1] & 0xffffff00) | (wx >> 24);
+ w1[2] = (w1[2] & 0x000000ff) | (wx << 8);
+ break;
+ case 24: w1[2] = wx;
+ break;
+ case 25: w1[2] = (w1[2] & 0xff000000) | (wx >> 8);
+ w1[3] = (w1[3] & 0x00ffffff) | (wx << 24);
+ break;
+ case 26: w1[2] = (w1[2] & 0xffff0000) | (wx >> 16);
+ w1[3] = (w1[3] & 0x0000ffff) | (wx << 16);
+ break;
+ case 27: w1[2] = (w1[2] & 0xffffff00) | (wx >> 24);
+ w1[3] = (w1[3] & 0x000000ff) | (wx << 8);
+ break;
+ case 28: w1[3] = wx;
+ break;
+ case 29: w1[3] = (w1[3] & 0xff000000) | (wx >> 8);
+ w2[0] = (w2[0] & 0x00ffffff) | (wx << 24);
+ break;
+ case 30: w1[3] = (w1[3] & 0xffff0000) | (wx >> 16);
+ w2[0] = (w2[0] & 0x0000ffff) | (wx << 16);
+ break;
+ case 31: w1[3] = (w1[3] & 0xffffff00) | (wx >> 24);
+ w2[0] = (w2[0] & 0x000000ff) | (wx << 8);
+ break;
+ case 32: w2[0] = wx;
+ break;
+ case 33: w2[0] = (w2[0] & 0xff000000) | (wx >> 8);
+ w2[1] = (w2[1] & 0x00ffffff) | (wx << 24);
+ break;
+ case 34: w2[0] = (w2[0] & 0xffff0000) | (wx >> 16);
+ w2[1] = (w2[1] & 0x0000ffff) | (wx << 16);
+ break;
+ case 35: w2[0] = (w2[0] & 0xffffff00) | (wx >> 24);
+ w2[1] = (w2[1] & 0x000000ff) | (wx << 8);
+ break;
+ case 36: w2[1] = wx;
+ break;
+ case 37: w2[1] = (w2[1] & 0xff000000) | (wx >> 8);
+ w2[2] = (w2[2] & 0x00ffffff) | (wx << 24);
+ break;
+ case 38: w2[1] = (w2[1] & 0xffff0000) | (wx >> 16);
+ w2[2] = (w2[2] & 0x0000ffff) | (wx << 16);
+ break;
+ case 39: w2[1] = (w2[1] & 0xffffff00) | (wx >> 24);
+ w2[2] = (w2[2] & 0x000000ff) | (wx << 8);
+ break;
+ case 40: w2[2] = wx;
+ break;
+ case 41: w2[2] = (w2[2] & 0xff000000) | (wx >> 8);
+ w2[3] = (w2[3] & 0x00ffffff) | (wx << 24);
+ break;
+ case 42: w2[2] = (w2[2] & 0xffff0000) | (wx >> 16);
+ w2[3] = (w2[3] & 0x0000ffff) | (wx << 16);
+ break;
+ case 43: w2[2] = (w2[2] & 0xffffff00) | (wx >> 24);
+ w2[3] = (w2[3] & 0x000000ff) | (wx << 8);
+ break;
+ case 44: w2[3] = wx;
+ break;
+ case 45: w2[3] = (w2[3] & 0xff000000) | (wx >> 8);
+ w3[0] = (w3[0] & 0x00ffffff) | (wx << 24);
+ break;
+ case 46: w2[3] = (w2[3] & 0xffff0000) | (wx >> 16);
+ w3[0] = (w3[0] & 0x0000ffff) | (wx << 16);
+ break;
+ case 47: w2[3] = (w2[3] & 0xffffff00) | (wx >> 24);
+ w3[0] = (w3[0] & 0x000000ff) | (wx << 8);
+ break;
+ case 48: w3[0] = wx;
+ break;
+ case 49: w3[0] = (w3[0] & 0xff000000) | (wx >> 8);
+ w3[1] = (w3[1] & 0x00ffffff) | (wx << 24);
+ break;
+ case 50: w3[0] = (w3[0] & 0xffff0000) | (wx >> 16);
+ w3[1] = (w3[1] & 0x0000ffff) | (wx << 16);
+ break;
+ case 51: w3[0] = (w3[0] & 0xffffff00) | (wx >> 24);
+ w3[1] = (w3[1] & 0x000000ff) | (wx << 8);
+ break;
+ case 52: w3[1] = wx;
+ break;
+ case 53: w3[1] = (w3[1] & 0xff000000) | (wx >> 8);
+ w3[2] = (w3[2] & 0x00ffffff) | (wx << 24);
+ break;
+ case 54: w3[1] = (w3[1] & 0xffff0000) | (wx >> 16);
+ w3[2] = (w3[2] & 0x0000ffff) | (wx << 16);
+ break;
+ case 55: w3[1] = (w3[1] & 0xffffff00) | (wx >> 24);
+ w3[2] = (w3[2] & 0x000000ff) | (wx << 8);
+ break;
+ case 56: w3[2] = wx;
+ break;
+ case 57: w3[2] = (w3[2] & 0xff000000) | (wx >> 8);
+ w3[3] = (w3[3] & 0x00ffffff) | (wx << 24);
+ break;
+ case 58: w3[2] = (w3[2] & 0xffff0000) | (wx >> 16);
+ w3[3] = (w3[3] & 0x0000ffff) | (wx << 16);
+ break;
+ case 59: w3[2] = (w3[2] & 0xffffff00) | (wx >> 24);
+ w3[3] = (w3[3] & 0x000000ff) | (wx << 8);
+ break;
+ case 60: w3[3] = wx;
+ break;
+ case 61: w3[3] = (w3[3] & 0xff000000) | (wx >> 8);
+ //w4[0] = (w4[0] & 0x00ffffff) | (wx << 24);
+ break;
+ case 62: w3[3] = (w3[3] & 0xffff0000) | (wx >> 16);
+ //w4[0] = (w4[0] & 0x0000ffff) | (wx << 16);
+ break;
+ case 63: w3[3] = (w3[3] & 0xffffff00) | (wx >> 24);
+ //w4[0] = (w4[0] & 0x000000ff) | (wx << 8);
+ break;
}
}
-// before: append_0x80_8
-static void append_0x80_1x32 (u32 w[32], const u32 offset)
+/**
+ * vector functions as scalar (for outer loop usage)
+ */
+
+inline void append_0x01_2x4_S (u32x w0[4], u32x w1[4], const u32 offset)
{
switch (offset)
{
case 0:
- w[ 0] = 0x80;
+ w0[0] = 0x01;
break;
case 1:
- w[ 0] = w[ 0] | 0x8000;
+ w0[0] = w0[0] | 0x0100;
break;
case 2:
- w[ 0] = w[ 0] | 0x800000;
+ w0[0] = w0[0] | 0x010000;
break;
case 3:
- w[ 0] = w[ 0] | 0x80000000;
+ w0[0] = w0[0] | 0x01000000;
break;
case 4:
- w[ 1] = 0x80;
+ w0[1] = 0x01;
break;
case 5:
- w[ 1] = w[ 1] | 0x8000;
+ w0[1] = w0[1] | 0x0100;
break;
case 6:
- w[ 1] = w[ 1] | 0x800000;
+ w0[1] = w0[1] | 0x010000;
break;
case 7:
- w[ 1] = w[ 1] | 0x80000000;
+ w0[1] = w0[1] | 0x01000000;
break;
case 8:
- w[ 2] = 0x80;
+ w0[2] = 0x01;
break;
case 9:
- w[ 2] = w[ 2] | 0x8000;
+ w0[2] = w0[2] | 0x0100;
break;
case 10:
- w[ 2] = w[ 2] | 0x800000;
+ w0[2] = w0[2] | 0x010000;
break;
case 11:
- w[ 2] = w[ 2] | 0x80000000;
+ w0[2] = w0[2] | 0x01000000;
break;
case 12:
- w[ 3] = 0x80;
+ w0[3] = 0x01;
break;
case 13:
- w[ 3] = w[ 3] | 0x8000;
+ w0[3] = w0[3] | 0x0100;
break;
case 14:
- w[ 3] = w[ 3] | 0x800000;
+ w0[3] = w0[3] | 0x010000;
break;
case 15:
- w[ 3] = w[ 3] | 0x80000000;
+ w0[3] = w0[3] | 0x01000000;
break;
case 16:
- w[ 4] = 0x80;
- break;
-
- case 17:
- w[ 4] = w[ 4] | 0x8000;
- break;
-
- case 18:
- w[ 4] = w[ 4] | 0x800000;
- break;
-
- case 19:
- w[ 4] = w[ 4] | 0x80000000;
- break;
-
- case 20:
- w[ 5] = 0x80;
- break;
-
- case 21:
- w[ 5] = w[ 5] | 0x8000;
- break;
-
- case 22:
- w[ 5] = w[ 5] | 0x800000;
- break;
-
- case 23:
- w[ 5] = w[ 5] | 0x80000000;
- break;
-
- case 24:
- w[ 6] = 0x80;
- break;
-
- case 25:
- w[ 6] = w[ 6] | 0x8000;
- break;
-
- case 26:
- w[ 6] = w[ 6] | 0x800000;
- break;
-
- case 27:
- w[ 6] = w[ 6] | 0x80000000;
- break;
-
- case 28:
- w[ 7] = 0x80;
- break;
-
- case 29:
- w[ 7] = w[ 7] | 0x8000;
- break;
-
- case 30:
- w[ 7] = w[ 7] | 0x800000;
- break;
-
- case 31:
- w[ 7] = w[ 7] | 0x80000000;
- break;
-
- case 32:
- w[ 8] = 0x80;
- break;
-
- case 33:
- w[ 8] = w[ 8] | 0x8000;
- break;
-
- case 34:
- w[ 8] = w[ 8] | 0x800000;
- break;
-
- case 35:
- w[ 8] = w[ 8] | 0x80000000;
- break;
-
- case 36:
- w[ 9] = 0x80;
- break;
-
- case 37:
- w[ 9] = w[ 9] | 0x8000;
- break;
-
- case 38:
- w[ 9] = w[ 9] | 0x800000;
- break;
-
- case 39:
- w[ 9] = w[ 9] | 0x80000000;
- break;
-
- case 40:
- w[10] = 0x80;
- break;
-
- case 41:
- w[10] = w[10] | 0x8000;
- break;
-
- case 42:
- w[10] = w[10] | 0x800000;
- break;
-
- case 43:
- w[10] = w[10] | 0x80000000;
- break;
-
- case 44:
- w[11] = 0x80;
- break;
-
- case 45:
- w[11] = w[11] | 0x8000;
- break;
-
- case 46:
- w[11] = w[11] | 0x800000;
- break;
-
- case 47:
- w[11] = w[11] | 0x80000000;
- break;
-
- case 48:
- w[12] = 0x80;
- break;
-
- case 49:
- w[12] = w[12] | 0x8000;
- break;
-
- case 50:
- w[12] = w[12] | 0x800000;
- break;
-
- case 51:
- w[12] = w[12] | 0x80000000;
- break;
-
- case 52:
- w[13] = 0x80;
- break;
-
- case 53:
- w[13] = w[13] | 0x8000;
- break;
-
- case 54:
- w[13] = w[13] | 0x800000;
- break;
-
- case 55:
- w[13] = w[13] | 0x80000000;
- break;
-
- case 56:
- w[14] = 0x80;
- break;
-
- case 57:
- w[14] = w[14] | 0x8000;
- break;
-
- case 58:
- w[14] = w[14] | 0x800000;
- break;
-
- case 59:
- w[14] = w[14] | 0x80000000;
- break;
-
- case 60:
- w[15] = 0x80;
- break;
-
- case 61:
- w[15] = w[15] | 0x8000;
- break;
-
- case 62:
- w[15] = w[15] | 0x800000;
- break;
-
- case 63:
- w[15] = w[15] | 0x80000000;
- break;
-
- case 64:
- w[16] = 0x80;
- break;
-
- case 65:
- w[16] = w[16] | 0x8000;
- break;
-
- case 66:
- w[16] = w[16] | 0x800000;
- break;
-
- case 67:
- w[16] = w[16] | 0x80000000;
- break;
-
- case 68:
- w[17] = 0x80;
- break;
-
- case 69:
- w[17] = w[17] | 0x8000;
- break;
-
- case 70:
- w[17] = w[17] | 0x800000;
- break;
-
- case 71:
- w[17] = w[17] | 0x80000000;
+ w1[0] = 0x01;
break;
- case 72:
- w[18] = 0x80;
+ case 17:
+ w1[0] = w1[0] | 0x0100;
break;
- case 73:
- w[18] = w[18] | 0x8000;
+ case 18:
+ w1[0] = w1[0] | 0x010000;
break;
- case 74:
- w[18] = w[18] | 0x800000;
+ case 19:
+ w1[0] = w1[0] | 0x01000000;
break;
- case 75:
- w[18] = w[18] | 0x80000000;
+ case 20:
+ w1[1] = 0x01;
break;
- case 76:
- w[19] = 0x80;
+ case 21:
+ w1[1] = w1[1] | 0x0100;
break;
- case 77:
- w[19] = w[19] | 0x8000;
+ case 22:
+ w1[1] = w1[1] | 0x010000;
break;
- case 78:
- w[19] = w[19] | 0x800000;
+ case 23:
+ w1[1] = w1[1] | 0x01000000;
break;
- case 79:
- w[19] = w[19] | 0x80000000;
+ case 24:
+ w1[2] = 0x01;
break;
- case 80:
- w[20] = 0x80;
+ case 25:
+ w1[2] = w1[2] | 0x0100;
break;
- case 81:
- w[20] = w[20] | 0x8000;
+ case 26:
+ w1[2] = w1[2] | 0x010000;
break;
- case 82:
- w[20] = w[20] | 0x800000;
+ case 27:
+ w1[2] = w1[2] | 0x01000000;
break;
- case 83:
- w[20] = w[20] | 0x80000000;
+ case 28:
+ w1[3] = 0x01;
break;
- case 84:
- w[21] = 0x80;
+ case 29:
+ w1[3] = w1[3] | 0x0100;
break;
- case 85:
- w[21] = w[21] | 0x8000;
+ case 30:
+ w1[3] = w1[3] | 0x010000;
break;
- case 86:
- w[21] = w[21] | 0x800000;
+ case 31:
+ w1[3] = w1[3] | 0x01000000;
break;
+ }
+}
- case 87:
- w[21] = w[21] | 0x80000000;
+inline void append_0x80_1x4_S (u32 w0[4], const u32 offset)
+{
+ switch (offset)
+ {
+ case 0:
+ w0[0] = 0x80;
break;
- case 88:
- w[22] = 0x80;
+ case 1:
+ w0[0] = w0[0] | 0x8000;
break;
- case 89:
- w[22] = w[22] | 0x8000;
+ case 2:
+ w0[0] = w0[0] | 0x800000;
break;
- case 90:
- w[22] = w[22] | 0x800000;
+ case 3:
+ w0[0] = w0[0] | 0x80000000;
break;
- case 91:
- w[22] = w[22] | 0x80000000;
+ case 4:
+ w0[1] = 0x80;
break;
- case 92:
- w[23] = 0x80;
+ case 5:
+ w0[1] = w0[1] | 0x8000;
break;
- case 93:
- w[23] = w[23] | 0x8000;
+ case 6:
+ w0[1] = w0[1] | 0x800000;
break;
- case 94:
- w[23] = w[23] | 0x800000;
+ case 7:
+ w0[1] = w0[1] | 0x80000000;
break;
- case 95:
- w[23] = w[23] | 0x80000000;
+ case 8:
+ w0[2] = 0x80;
break;
- case 96:
- w[24] = 0x80;
+ case 9:
+ w0[2] = w0[2] | 0x8000;
break;
- case 97:
- w[24] = w[24] | 0x8000;
+ case 10:
+ w0[2] = w0[2] | 0x800000;
break;
- case 98:
- w[24] = w[24] | 0x800000;
+ case 11:
+ w0[2] = w0[2] | 0x80000000;
break;
- case 99:
- w[24] = w[24] | 0x80000000;
+ case 12:
+ w0[3] = 0x80;
break;
- case 100:
- w[25] = 0x80;
+ case 13:
+ w0[3] = w0[3] | 0x8000;
break;
- case 101:
- w[25] = w[25] | 0x8000;
+ case 14:
+ w0[3] = w0[3] | 0x800000;
break;
- case 102:
- w[25] = w[25] | 0x800000;
+ case 15:
+ w0[3] = w0[3] | 0x80000000;
break;
+ }
+}
- case 103:
- w[25] = w[25] | 0x80000000;
+inline void append_0x80_2x4_S (u32 w0[4], u32 w1[4], const u32 offset)
+{
+ switch (offset)
+ {
+ case 0:
+ w0[0] = 0x80;
break;
- case 104:
- w[26] = 0x80;
+ case 1:
+ w0[0] = w0[0] | 0x8000;
break;
- case 105:
- w[26] = w[26] | 0x8000;
+ case 2:
+ w0[0] = w0[0] | 0x800000;
break;
- case 106:
- w[26] = w[26] | 0x800000;
+ case 3:
+ w0[0] = w0[0] | 0x80000000;
break;
- case 107:
- w[26] = w[26] | 0x80000000;
+ case 4:
+ w0[1] = 0x80;
break;
- case 108:
- w[27] = 0x80;
+ case 5:
+ w0[1] = w0[1] | 0x8000;
break;
- case 109:
- w[27] = w[27] | 0x8000;
+ case 6:
+ w0[1] = w0[1] | 0x800000;
break;
- case 110:
- w[27] = w[27] | 0x800000;
+ case 7:
+ w0[1] = w0[1] | 0x80000000;
break;
- case 111:
- w[27] = w[27] | 0x80000000;
+ case 8:
+ w0[2] = 0x80;
break;
- case 112:
- w[28] = 0x80;
+ case 9:
+ w0[2] = w0[2] | 0x8000;
break;
- case 113:
- w[28] = w[28] | 0x8000;
+ case 10:
+ w0[2] = w0[2] | 0x800000;
break;
- case 114:
- w[28] = w[28] | 0x800000;
+ case 11:
+ w0[2] = w0[2] | 0x80000000;
break;
- case 115:
- w[28] = w[28] | 0x80000000;
+ case 12:
+ w0[3] = 0x80;
break;
- case 116:
- w[29] = 0x80;
+ case 13:
+ w0[3] = w0[3] | 0x8000;
break;
- case 117:
- w[29] = w[29] | 0x8000;
+ case 14:
+ w0[3] = w0[3] | 0x800000;
break;
- case 118:
- w[29] = w[29] | 0x800000;
+ case 15:
+ w0[3] = w0[3] | 0x80000000;
break;
- case 119:
- w[29] = w[29] | 0x80000000;
+ case 16:
+ w1[0] = 0x80;
break;
- case 120:
- w[30] = 0x80;
+ case 17:
+ w1[0] = w1[0] | 0x8000;
break;
- case 121:
- w[30] = w[30] | 0x8000;
+ case 18:
+ w1[0] = w1[0] | 0x800000;
break;
- case 122:
- w[30] = w[30] | 0x800000;
+ case 19:
+ w1[0] = w1[0] | 0x80000000;
break;
- case 123:
- w[30] = w[30] | 0x80000000;
+ case 20:
+ w1[1] = 0x80;
break;
- case 124:
- w[31] = 0x80;
+ case 21:
+ w1[1] = w1[1] | 0x8000;
break;
- case 125:
- w[31] = w[31] | 0x8000;
+ case 22:
+ w1[1] = w1[1] | 0x800000;
break;
- case 126:
- w[31] = w[31] | 0x800000;
+ case 23:
+ w1[1] = w1[1] | 0x80000000;
break;
- case 127:
- w[31] = w[31] | 0x80000000;
+ case 24:
+ w1[2] = 0x80;
break;
- }
-}
-// before: device_memcat2L
-static void memcat_c7_d1x2_sl1x2_sr1x2 (const u32 offset, u32 dst0[2], u32 src_l0[2], u32 src_r0[2])
-{
- switch (offset)
- {
- case 1:
- dst0[0] = src_l0[0] | src_r0[0] << 8;
- dst0[1] = src_r0[0] >> 24 | src_r0[1] << 8;
+ case 25:
+ w1[2] = w1[2] | 0x8000;
break;
- case 2:
- dst0[0] = src_l0[0] | src_r0[0] << 16;
- dst0[1] = src_r0[0] >> 16 | src_r0[1] << 16;
+ case 26:
+ w1[2] = w1[2] | 0x800000;
break;
- case 3:
- dst0[0] = src_l0[0] | src_r0[0] << 24;
- dst0[1] = src_r0[0] >> 8 | src_r0[1] << 24;
+ case 27:
+ w1[2] = w1[2] | 0x80000000;
break;
- case 4:
- dst0[1] = src_r0[0];
+ case 28:
+ w1[3] = 0x80;
break;
- case 5:
- dst0[1] = src_l0[1] | src_r0[0] << 8;
+ case 29:
+ w1[3] = w1[3] | 0x8000;
break;
- case 6:
- dst0[1] = src_l0[1] | src_r0[0] << 16;
+ case 30:
+ w1[3] = w1[3] | 0x800000;
break;
- case 7:
- dst0[1] = src_l0[1] | src_r0[0] << 24;
+ case 31:
+ w1[3] = w1[3] | 0x80000000;
break;
}
}
-// before: device_memcat4L
-static void memcat_c15_d1x4_sl1x4_sr1x4 (const u32 offset, u32 dst0[4], u32 src_l0[4], u32 src_r0[4])
+inline void append_0x80_3x4_S (u32 w0[4], u32 w1[4], u32 w2[4], const u32 offset)
{
switch (offset)
{
+ case 0:
+ w0[0] = 0x80;
+ break;
+
case 1:
- dst0[0] = src_l0[0] | src_r0[0] << 8;
- dst0[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst0[3] = src_r0[2] >> 24 | src_r0[3] << 8;
+ w0[0] = w0[0] | 0x8000;
break;
case 2:
- dst0[0] = src_l0[0] | src_r0[0] << 16;
- dst0[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst0[3] = src_r0[2] >> 16 | src_r0[3] << 16;
+ w0[0] = w0[0] | 0x800000;
break;
case 3:
- dst0[0] = src_l0[0] | src_r0[0] << 24;
- dst0[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst0[3] = src_r0[2] >> 8 | src_r0[3] << 24;
+ w0[0] = w0[0] | 0x80000000;
break;
case 4:
- dst0[1] = src_r0[0];
- dst0[2] = src_r0[1];
- dst0[3] = src_r0[2];
+ w0[1] = 0x80;
break;
case 5:
- dst0[1] = src_l0[1] | src_r0[0] << 8;
- dst0[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[3] = src_r0[1] >> 24 | src_r0[2] << 8;
+ w0[1] = w0[1] | 0x8000;
break;
case 6:
- dst0[1] = src_l0[1] | src_r0[0] << 16;
- dst0[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[3] = src_r0[1] >> 16 | src_r0[2] << 16;
+ w0[1] = w0[1] | 0x800000;
break;
case 7:
- dst0[1] = src_l0[1] | src_r0[0] << 24;
- dst0[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[3] = src_r0[1] >> 8 | src_r0[2] << 24;
+ w0[1] = w0[1] | 0x80000000;
break;
case 8:
- dst0[2] = src_r0[0];
- dst0[3] = src_r0[1];
+ w0[2] = 0x80;
break;
case 9:
- dst0[2] = src_l0[2] | src_r0[0] << 8;
- dst0[3] = src_r0[0] >> 24 | src_r0[1] << 8;
+ w0[2] = w0[2] | 0x8000;
break;
case 10:
- dst0[2] = src_l0[2] | src_r0[0] << 16;
- dst0[3] = src_r0[0] >> 16 | src_r0[1] << 16;
+ w0[2] = w0[2] | 0x800000;
break;
case 11:
- dst0[2] = src_l0[2] | src_r0[0] << 24;
- dst0[3] = src_r0[0] >> 8 | src_r0[1] << 24;
+ w0[2] = w0[2] | 0x80000000;
break;
case 12:
- dst0[3] = src_r0[0];
+ w0[3] = 0x80;
break;
case 13:
- dst0[3] = src_l0[3] | src_r0[0] << 8;
+ w0[3] = w0[3] | 0x8000;
break;
case 14:
- dst0[3] = src_l0[3] | src_r0[0] << 16;
+ w0[3] = w0[3] | 0x800000;
break;
case 15:
- dst0[3] = src_l0[3] | src_r0[0] << 24;
+ w0[3] = w0[3] | 0x80000000;
break;
- }
-}
-// before: device_memcat8L
-static void memcat_c31_d2x4_sl2x4_sr1x4 (const u32 offset, u32 dst0[4], u32 dst1[4], u32 src_l0[4], u32 src_l1[4], u32 src_r0[4])
-{
- switch (offset)
- {
- case 1:
- dst0[0] = src_l0[0] | src_r0[0] << 8;
- dst0[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst0[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[0] = src_r0[3] >> 24;
+ case 16:
+ w1[0] = 0x80;
break;
- case 2:
- dst0[0] = src_l0[0] | src_r0[0] << 16;
- dst0[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst0[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[0] = src_r0[3] >> 16;
+ case 17:
+ w1[0] = w1[0] | 0x8000;
break;
- case 3:
- dst0[0] = src_l0[0] | src_r0[0] << 24;
- dst0[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst0[3] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[0] = src_r0[3] >> 8;
+ case 18:
+ w1[0] = w1[0] | 0x800000;
break;
- case 4:
- dst0[1] = src_r0[0];
- dst0[2] = src_r0[1];
- dst0[3] = src_r0[2];
- dst1[0] = src_r0[3];
+ case 19:
+ w1[0] = w1[0] | 0x80000000;
break;
- case 5:
- dst0[1] = src_l0[1] | src_r0[0] << 8;
- dst0[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[0] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[1] = src_r0[3] >> 24;
+ case 20:
+ w1[1] = 0x80;
break;
- case 6:
- dst0[1] = src_l0[1] | src_r0[0] << 16;
- dst0[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[3] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[0] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[1] = src_r0[3] >> 16;
+ case 21:
+ w1[1] = w1[1] | 0x8000;
break;
- case 7:
- dst0[1] = src_l0[1] | src_r0[0] << 24;
- dst0[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[0] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[1] = src_r0[3] >> 8;
+ case 22:
+ w1[1] = w1[1] | 0x800000;
break;
- case 8:
- dst0[2] = src_r0[0];
- dst0[3] = src_r0[1];
- dst1[0] = src_r0[2];
- dst1[1] = src_r0[3];
+ case 23:
+ w1[1] = w1[1] | 0x80000000;
break;
- case 9:
- dst0[2] = src_l0[2] | src_r0[0] << 8;
- dst0[3] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[0] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[1] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[2] = src_r0[3] >> 24;
+ case 24:
+ w1[2] = 0x80;
+ break;
+
+ case 25:
+ w1[2] = w1[2] | 0x8000;
break;
- case 10:
- dst0[2] = src_l0[2] | src_r0[0] << 16;
- dst0[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[0] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[1] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[2] = src_r0[3] >> 16;
+ case 26:
+ w1[2] = w1[2] | 0x800000;
break;
- case 11:
- dst0[2] = src_l0[2] | src_r0[0] << 24;
- dst0[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[0] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[1] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[2] = src_r0[3] >> 8;
+ case 27:
+ w1[2] = w1[2] | 0x80000000;
break;
- case 12:
- dst0[3] = src_r0[0];
- dst1[0] = src_r0[1];
- dst1[1] = src_r0[2];
- dst1[2] = src_r0[3];
+ case 28:
+ w1[3] = 0x80;
break;
- case 13:
- dst0[3] = src_l0[3] | src_r0[0] << 8;
- dst1[0] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[1] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[2] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[3] = src_r0[3] >> 24;
+ case 29:
+ w1[3] = w1[3] | 0x8000;
break;
- case 14:
- dst0[3] = src_l0[3] | src_r0[0] << 16;
- dst1[0] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[1] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[2] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[3] = src_r0[3] >> 16;
+ case 30:
+ w1[3] = w1[3] | 0x800000;
break;
- case 15:
- dst0[3] = src_l0[3] | src_r0[0] << 24;
- dst1[0] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[1] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[2] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[3] = src_r0[3] >> 8;
+ case 31:
+ w1[3] = w1[3] | 0x80000000;
break;
- case 16:
- dst1[0] = src_r0[0];
- dst1[1] = src_r0[1];
- dst1[2] = src_r0[2];
- dst1[3] = src_r0[3];
+ case 32:
+ w2[0] = 0x80;
break;
- case 17:
- dst1[0] = src_l1[0] | src_r0[0] << 8;
- dst1[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[3] = src_r0[2] >> 24 | src_r0[3] << 8;
+ case 33:
+ w2[0] = w2[0] | 0x8000;
break;
- case 18:
- dst1[0] = src_l1[0] | src_r0[0] << 16;
- dst1[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[3] = src_r0[2] >> 16 | src_r0[3] << 16;
+ case 34:
+ w2[0] = w2[0] | 0x800000;
break;
- case 19:
- dst1[0] = src_l1[0] | src_r0[0] << 24;
- dst1[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[3] = src_r0[2] >> 8 | src_r0[3] << 24;
+ case 35:
+ w2[0] = w2[0] | 0x80000000;
break;
- case 20:
- dst1[1] = src_r0[0];
- dst1[2] = src_r0[1];
- dst1[3] = src_r0[2];
+ case 36:
+ w2[1] = 0x80;
break;
- case 21:
- dst1[1] = src_l1[1] | src_r0[0] << 8;
- dst1[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[3] = src_r0[1] >> 24 | src_r0[2] << 8;
+ case 37:
+ w2[1] = w2[1] | 0x8000;
break;
- case 22:
- dst1[1] = src_l1[1] | src_r0[0] << 16;
- dst1[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[3] = src_r0[1] >> 16 | src_r0[2] << 16;
+ case 38:
+ w2[1] = w2[1] | 0x800000;
break;
- case 23:
- dst1[1] = src_l1[1] | src_r0[0] << 24;
- dst1[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[3] = src_r0[1] >> 8 | src_r0[2] << 24;
+ case 39:
+ w2[1] = w2[1] | 0x80000000;
break;
- case 24:
- dst1[2] = src_r0[0];
- dst1[3] = src_r0[1];
+ case 40:
+ w2[2] = 0x80;
break;
- case 25:
- dst1[2] = src_l1[2] | src_r0[0] << 8;
- dst1[3] = src_r0[0] >> 24 | src_r0[1] << 8;
+ case 41:
+ w2[2] = w2[2] | 0x8000;
break;
- case 26:
- dst1[2] = src_l1[2] | src_r0[0] << 16;
- dst1[3] = src_r0[0] >> 16 | src_r0[1] << 16;
+ case 42:
+ w2[2] = w2[2] | 0x800000;
break;
- case 27:
- dst1[2] = src_l1[2] | src_r0[0] << 24;
- dst1[3] = src_r0[0] >> 8 | src_r0[1] << 24;
+ case 43:
+ w2[2] = w2[2] | 0x80000000;
break;
- case 28:
- dst1[3] = src_r0[0];
+ case 44:
+ w2[3] = 0x80;
break;
- case 29:
- dst1[3] = src_l1[3] | src_r0[0] << 8;
+ case 45:
+ w2[3] = w2[3] | 0x8000;
break;
- case 30:
- dst1[3] = src_l1[3] | src_r0[0] << 16;
+ case 46:
+ w2[3] = w2[3] | 0x800000;
break;
- case 31:
- dst1[3] = src_l1[3] | src_r0[0] << 24;
+ case 47:
+ w2[3] = w2[3] | 0x80000000;
break;
}
}
-// before: device_memcat12L
-static void memcat_c47_d3x4_sl3x4_sr1x4 (const u32 offset, u32 dst0[4], u32 dst1[4], u32 dst2[4], u32 src_l0[4], u32 src_l1[4], u32 src_l2[4], u32 src_r0[4])
+inline void append_0x80_4x4_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
{
switch (offset)
{
+ case 0:
+ w0[0] = 0x80;
+ break;
+
case 1:
- dst0[0] = src_l0[0] | src_r0[0] << 8;
- dst0[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst0[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[0] = src_r0[3] >> 24;
+ w0[0] = w0[0] | 0x8000;
break;
case 2:
- dst0[0] = src_l0[0] | src_r0[0] << 16;
- dst0[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst0[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[0] = src_r0[3] >> 16;
+ w0[0] = w0[0] | 0x800000;
break;
case 3:
- dst0[0] = src_l0[0] | src_r0[0] << 24;
- dst0[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst0[3] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[0] = src_r0[3] >> 8;
+ w0[0] = w0[0] | 0x80000000;
break;
case 4:
- dst0[1] = src_r0[0];
- dst0[2] = src_r0[1];
- dst0[3] = src_r0[2];
- dst1[0] = src_r0[3];
+ w0[1] = 0x80;
break;
case 5:
- dst0[1] = src_l0[1] | src_r0[0] << 8;
- dst0[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[0] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[1] = src_r0[3] >> 24;
+ w0[1] = w0[1] | 0x8000;
break;
case 6:
- dst0[1] = src_l0[1] | src_r0[0] << 16;
- dst0[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[3] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[0] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[1] = src_r0[3] >> 16;
+ w0[1] = w0[1] | 0x800000;
break;
case 7:
- dst0[1] = src_l0[1] | src_r0[0] << 24;
- dst0[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[0] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[1] = src_r0[3] >> 8;
+ w0[1] = w0[1] | 0x80000000;
break;
case 8:
- dst0[2] = src_r0[0];
- dst0[3] = src_r0[1];
- dst1[0] = src_r0[2];
- dst1[1] = src_r0[3];
+ w0[2] = 0x80;
break;
case 9:
- dst0[2] = src_l0[2] | src_r0[0] << 8;
- dst0[3] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[0] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[1] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[2] = src_r0[3] >> 24;
+ w0[2] = w0[2] | 0x8000;
break;
case 10:
- dst0[2] = src_l0[2] | src_r0[0] << 16;
- dst0[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[0] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[1] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[2] = src_r0[3] >> 16;
+ w0[2] = w0[2] | 0x800000;
break;
case 11:
- dst0[2] = src_l0[2] | src_r0[0] << 24;
- dst0[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[0] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[1] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[2] = src_r0[3] >> 8;
+ w0[2] = w0[2] | 0x80000000;
break;
case 12:
- dst0[3] = src_r0[0];
- dst1[0] = src_r0[1];
- dst1[1] = src_r0[2];
- dst1[2] = src_r0[3];
+ w0[3] = 0x80;
break;
case 13:
- dst0[3] = src_l0[3] | src_r0[0] << 8;
- dst1[0] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[1] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[2] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[3] = src_r0[3] >> 24;
+ w0[3] = w0[3] | 0x8000;
break;
case 14:
- dst0[3] = src_l0[3] | src_r0[0] << 16;
- dst1[0] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[1] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[2] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[3] = src_r0[3] >> 16;
+ w0[3] = w0[3] | 0x800000;
break;
case 15:
- dst0[3] = src_l0[3] | src_r0[0] << 24;
- dst1[0] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[1] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[2] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[3] = src_r0[3] >> 8;
+ w0[3] = w0[3] | 0x80000000;
break;
case 16:
- dst1[0] = src_r0[0];
- dst1[1] = src_r0[1];
- dst1[2] = src_r0[2];
- dst1[3] = src_r0[3];
+ w1[0] = 0x80;
break;
case 17:
- dst1[0] = src_l1[0] | src_r0[0] << 8;
- dst1[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[0] = src_r0[3] >> 24;
+ w1[0] = w1[0] | 0x8000;
break;
case 18:
- dst1[0] = src_l1[0] | src_r0[0] << 16;
- dst1[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[0] = src_r0[3] >> 16;
+ w1[0] = w1[0] | 0x800000;
break;
case 19:
- dst1[0] = src_l1[0] | src_r0[0] << 24;
- dst1[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[3] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[0] = src_r0[3] >> 8;
+ w1[0] = w1[0] | 0x80000000;
break;
case 20:
- dst1[1] = src_r0[0];
- dst1[2] = src_r0[1];
- dst1[3] = src_r0[2];
- dst2[0] = src_r0[3];
+ w1[1] = 0x80;
break;
case 21:
- dst1[1] = src_l1[1] | src_r0[0] << 8;
- dst1[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[0] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[1] = src_r0[3] >> 24;
+ w1[1] = w1[1] | 0x8000;
break;
case 22:
- dst1[1] = src_l1[1] | src_r0[0] << 16;
- dst1[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[3] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[0] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[1] = src_r0[3] >> 16;
+ w1[1] = w1[1] | 0x800000;
break;
case 23:
- dst1[1] = src_l1[1] | src_r0[0] << 24;
- dst1[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[0] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[1] = src_r0[3] >> 8;
+ w1[1] = w1[1] | 0x80000000;
break;
case 24:
- dst1[2] = src_r0[0];
- dst1[3] = src_r0[1];
- dst2[0] = src_r0[2];
- dst2[1] = src_r0[3];
+ w1[2] = 0x80;
break;
case 25:
- dst1[2] = src_l1[2] | src_r0[0] << 8;
- dst1[3] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[0] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[1] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[2] = src_r0[3] >> 24;
+ w1[2] = w1[2] | 0x8000;
break;
case 26:
- dst1[2] = src_l1[2] | src_r0[0] << 16;
- dst1[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[0] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[1] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[2] = src_r0[3] >> 16;
+ w1[2] = w1[2] | 0x800000;
break;
case 27:
- dst1[2] = src_l1[2] | src_r0[0] << 24;
- dst1[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[0] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[1] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[2] = src_r0[3] >> 8;
+ w1[2] = w1[2] | 0x80000000;
break;
case 28:
- dst1[3] = src_r0[0];
- dst2[0] = src_r0[1];
- dst2[1] = src_r0[2];
- dst2[2] = src_r0[3];
+ w1[3] = 0x80;
break;
case 29:
- dst1[3] = src_l1[3] | src_r0[0] << 8;
- dst2[0] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[1] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[2] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[3] = src_r0[3] >> 24;
+ w1[3] = w1[3] | 0x8000;
break;
case 30:
- dst1[3] = src_l1[3] | src_r0[0] << 16;
- dst2[0] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[1] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[2] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[3] = src_r0[3] >> 16;
+ w1[3] = w1[3] | 0x800000;
break;
case 31:
- dst1[3] = src_l1[3] | src_r0[0] << 24;
- dst2[0] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[1] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[2] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[3] = src_r0[3] >> 8;
+ w1[3] = w1[3] | 0x80000000;
break;
case 32:
- dst2[0] = src_r0[0];
- dst2[1] = src_r0[1];
- dst2[2] = src_r0[2];
- dst2[3] = src_r0[3];
+ w2[0] = 0x80;
break;
case 33:
- dst2[0] = src_l2[0] | src_r0[0] << 8;
- dst2[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[3] = src_r0[2] >> 24 | src_r0[3] << 8;
+ w2[0] = w2[0] | 0x8000;
break;
case 34:
- dst2[0] = src_l2[0] | src_r0[0] << 16;
- dst2[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[3] = src_r0[2] >> 16 | src_r0[3] << 16;
+ w2[0] = w2[0] | 0x800000;
break;
case 35:
- dst2[0] = src_l2[0] | src_r0[0] << 24;
- dst2[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[3] = src_r0[2] >> 8 | src_r0[3] << 24;
+ w2[0] = w2[0] | 0x80000000;
break;
case 36:
- dst2[1] = src_r0[0];
- dst2[2] = src_r0[1];
- dst2[3] = src_r0[2];
+ w2[1] = 0x80;
break;
case 37:
- dst2[1] = src_l2[1] | src_r0[0] << 8;
- dst2[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[3] = src_r0[1] >> 24 | src_r0[2] << 8;
+ w2[1] = w2[1] | 0x8000;
break;
case 38:
- dst2[1] = src_l2[1] | src_r0[0] << 16;
- dst2[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[3] = src_r0[1] >> 16 | src_r0[2] << 16;
+ w2[1] = w2[1] | 0x800000;
break;
case 39:
- dst2[1] = src_l2[1] | src_r0[0] << 24;
- dst2[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[3] = src_r0[1] >> 8 | src_r0[2] << 24;
+ w2[1] = w2[1] | 0x80000000;
break;
case 40:
- dst2[2] = src_r0[0];
- dst2[3] = src_r0[1];
+ w2[2] = 0x80;
break;
case 41:
- dst2[2] = src_l2[2] | src_r0[0] << 8;
- dst2[3] = src_r0[0] >> 24 | src_r0[1] << 8;
+ w2[2] = w2[2] | 0x8000;
break;
case 42:
- dst2[2] = src_l2[2] | src_r0[0] << 16;
- dst2[3] = src_r0[0] >> 16 | src_r0[1] << 16;
+ w2[2] = w2[2] | 0x800000;
break;
case 43:
- dst2[2] = src_l2[2] | src_r0[0] << 24;
- dst2[3] = src_r0[0] >> 8 | src_r0[1] << 24;
+ w2[2] = w2[2] | 0x80000000;
break;
case 44:
- dst2[3] = src_r0[0];
+ w2[3] = 0x80;
break;
case 45:
- dst2[3] = src_l2[3] | src_r0[0] << 8;
+ w2[3] = w2[3] | 0x8000;
break;
case 46:
- dst2[3] = src_l2[3] | src_r0[0] << 16;
+ w2[3] = w2[3] | 0x800000;
+ break;
+
+ case 47:
+ w2[3] = w2[3] | 0x80000000;
+ break;
+
+ case 48:
+ w3[0] = 0x80;
+ break;
+
+ case 49:
+ w3[0] = w3[0] | 0x8000;
+ break;
+
+ case 50:
+ w3[0] = w3[0] | 0x800000;
+ break;
+
+ case 51:
+ w3[0] = w3[0] | 0x80000000;
+ break;
+
+ case 52:
+ w3[1] = 0x80;
+ break;
+
+ case 53:
+ w3[1] = w3[1] | 0x8000;
+ break;
+
+ case 54:
+ w3[1] = w3[1] | 0x800000;
+ break;
+
+ case 55:
+ w3[1] = w3[1] | 0x80000000;
+ break;
+
+ case 56:
+ w3[2] = 0x80;
+ break;
+
+ case 57:
+ w3[2] = w3[2] | 0x8000;
+ break;
+
+ case 58:
+ w3[2] = w3[2] | 0x800000;
+ break;
+
+ case 59:
+ w3[2] = w3[2] | 0x80000000;
+ break;
+
+ case 60:
+ w3[3] = 0x80;
+ break;
+
+ case 61:
+ w3[3] = w3[3] | 0x8000;
+ break;
+
+ case 62:
+ w3[3] = w3[3] | 0x800000;
+ break;
+
+ case 63:
+ w3[3] = w3[3] | 0x80000000;
break;
+ }
+}
+
+inline void truncate_block_S (u32 w[4], const u32 len)
+{
+ switch (len)
+ {
+ case 0: w[0] &= 0;
+ w[1] &= 0;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 1: w[0] &= 0x000000FF;
+ w[1] &= 0;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 2: w[0] &= 0x0000FFFF;
+ w[1] &= 0;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 3: w[0] &= 0x00FFFFFF;
+ w[1] &= 0;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 4: w[1] &= 0;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 5: w[1] &= 0x000000FF;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 6: w[1] &= 0x0000FFFF;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 7: w[1] &= 0x00FFFFFF;
+ w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 8: w[2] &= 0;
+ w[3] &= 0;
+ break;
+ case 9: w[2] &= 0x000000FF;
+ w[3] &= 0;
+ break;
+ case 10: w[2] &= 0x0000FFFF;
+ w[3] &= 0;
+ break;
+ case 11: w[2] &= 0x00FFFFFF;
+ w[3] &= 0;
+ break;
+ case 12: w[3] &= 0;
+ break;
+ case 13: w[3] &= 0x000000FF;
+ break;
+ case 14: w[3] &= 0x0000FFFF;
+ break;
+ case 15: w[3] &= 0x00FFFFFF;
+ break;
+ }
+}
+
+inline void make_unicode_S (const u32 in[4], u32 out1[4], u32 out2[4])
+{
+ #ifdef IS_NV
+ out2[3] = __byte_perm_S (in[3], 0, 0x7372);
+ out2[2] = __byte_perm_S (in[3], 0, 0x7170);
+ out2[1] = __byte_perm_S (in[2], 0, 0x7372);
+ out2[0] = __byte_perm_S (in[2], 0, 0x7170);
+ out1[3] = __byte_perm_S (in[1], 0, 0x7372);
+ out1[2] = __byte_perm_S (in[1], 0, 0x7170);
+ out1[1] = __byte_perm_S (in[0], 0, 0x7372);
+ out1[0] = __byte_perm_S (in[0], 0, 0x7170);
+ #endif
- case 47:
- dst2[3] = src_l2[3] | src_r0[0] << 24;
- break;
- }
+ #if defined IS_AMD || defined IS_GENERIC
+ out2[3] = ((in[3] >> 8) & 0x00FF0000) | ((in[3] >> 16) & 0x000000FF);
+ out2[2] = ((in[3] << 8) & 0x00FF0000) | ((in[3] >> 0) & 0x000000FF);
+ out2[1] = ((in[2] >> 8) & 0x00FF0000) | ((in[2] >> 16) & 0x000000FF);
+ out2[0] = ((in[2] << 8) & 0x00FF0000) | ((in[2] >> 0) & 0x000000FF);
+ out1[3] = ((in[1] >> 8) & 0x00FF0000) | ((in[1] >> 16) & 0x000000FF);
+ out1[2] = ((in[1] << 8) & 0x00FF0000) | ((in[1] >> 0) & 0x000000FF);
+ out1[1] = ((in[0] >> 8) & 0x00FF0000) | ((in[0] >> 16) & 0x000000FF);
+ out1[0] = ((in[0] << 8) & 0x00FF0000) | ((in[0] >> 0) & 0x000000FF);
+ #endif
}
-// before: device_memcat12L
-static void memcat_c47_d3x4_sl3x4_sr2x4 (const u32 offset, u32 dst0[4], u32 dst1[4], u32 dst2[4], u32 src_l0[4], u32 src_l1[4], u32 src_l2[4], u32 src_r0[4], u32 src_r1[4])
+inline void undo_unicode_S (const u32 in1[4], const u32 in2[4], u32 out[4])
{
- switch (offset)
- {
- case 0:
- dst0[0] = src_r0[0];
- dst0[1] = src_r0[1];
- dst0[2] = src_r0[2];
- dst0[3] = src_r0[3];
- dst1[0] = src_r1[0];
- dst1[1] = src_r1[1];
- dst1[2] = src_r1[2];
- dst1[3] = src_r1[3];
- break;
-
- case 1:
- dst0[0] = src_l0[0] | src_r0[0] << 8;
- dst0[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst0[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[0] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst1[1] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst1[2] = src_r1[1] >> 24 | src_r1[2] << 8;
- dst1[3] = src_r1[2] >> 24 | src_r1[3] << 8;
- dst2[0] = src_r1[3] >> 24;
- break;
-
- case 2:
- dst0[0] = src_l0[0] | src_r0[0] << 16;
- dst0[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst0[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[0] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst1[1] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst1[2] = src_r1[1] >> 16 | src_r1[2] << 16;
- dst1[3] = src_r1[2] >> 16 | src_r1[3] << 16;
- dst2[0] = src_r1[3] >> 16;
- break;
+ #ifdef IS_NV
+ out[0] = __byte_perm_S (in1[0], in1[1], 0x6420);
+ out[1] = __byte_perm_S (in1[2], in1[3], 0x6420);
+ out[2] = __byte_perm_S (in2[0], in2[1], 0x6420);
+ out[3] = __byte_perm_S (in2[2], in2[3], 0x6420);
+ #endif
- case 3:
- dst0[0] = src_l0[0] | src_r0[0] << 24;
- dst0[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst0[3] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[0] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst1[1] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst1[2] = src_r1[1] >> 8 | src_r1[2] << 24;
- dst1[3] = src_r1[2] >> 8 | src_r1[3] << 24;
- dst2[0] = src_r1[3] >> 8;
- break;
+ #if defined IS_AMD || defined IS_GENERIC
+ out[0] = ((in1[0] & 0x000000ff) >> 0) | ((in1[0] & 0x00ff0000) >> 8)
+ | ((in1[1] & 0x000000ff) << 16) | ((in1[1] & 0x00ff0000) << 8);
+ out[1] = ((in1[2] & 0x000000ff) >> 0) | ((in1[2] & 0x00ff0000) >> 8)
+ | ((in1[3] & 0x000000ff) << 16) | ((in1[3] & 0x00ff0000) << 8);
+ out[2] = ((in2[0] & 0x000000ff) >> 0) | ((in2[0] & 0x00ff0000) >> 8)
+ | ((in2[1] & 0x000000ff) << 16) | ((in2[1] & 0x00ff0000) << 8);
+ out[3] = ((in2[2] & 0x000000ff) >> 0) | ((in2[2] & 0x00ff0000) >> 8)
+ | ((in2[3] & 0x000000ff) << 16) | ((in2[3] & 0x00ff0000) << 8);
+ #endif
+}
- case 4:
- dst0[1] = src_r0[0];
- dst0[2] = src_r0[1];
- dst0[3] = src_r0[2];
- dst1[0] = src_r0[3];
- dst1[1] = src_r1[0];
- dst1[2] = src_r1[1];
- dst1[3] = src_r1[2];
- dst2[0] = src_r1[3];
- break;
+inline void switch_buffer_by_offset_le_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
+{
+ #if defined IS_AMD || defined IS_GENERIC
+ const int offset_mod_4 = offset & 3;
- case 5:
- dst0[1] = src_l0[1] | src_r0[0] << 8;
- dst0[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst0[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[0] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[1] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst1[2] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst1[3] = src_r1[1] >> 24 | src_r1[2] << 8;
- dst2[0] = src_r1[2] >> 24 | src_r1[3] << 8;
- dst2[1] = src_r1[3] >> 24;
- break;
+ const int offset_minus_4 = 4 - offset;
- case 6:
- dst0[1] = src_l0[1] | src_r0[0] << 16;
- dst0[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst0[3] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[0] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[1] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst1[2] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst1[3] = src_r1[1] >> 16 | src_r1[2] << 16;
- dst2[0] = src_r1[2] >> 16 | src_r1[3] << 16;
- dst2[1] = src_r1[3] >> 16;
- break;
+ switch (offset / 4)
+ {
+ case 0:
+ w3[2] = amd_bytealign_S ( 0, w3[1], offset_minus_4);
+ w3[1] = amd_bytealign_S (w3[1], w3[0], offset_minus_4);
+ w3[0] = amd_bytealign_S (w3[0], w2[3], offset_minus_4);
+ w2[3] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
+ w2[2] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
+ w2[1] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
+ w2[0] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w1[3] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w1[2] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w1[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w1[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w0[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w0[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w0[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w0[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
- case 7:
- dst0[1] = src_l0[1] | src_r0[0] << 24;
- dst0[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst0[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[0] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[1] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst1[2] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst1[3] = src_r1[1] >> 8 | src_r1[2] << 24;
- dst2[0] = src_r1[2] >> 8 | src_r1[3] << 24;
- dst2[1] = src_r1[3] >> 8;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w0[0] = w0[1];
+ w0[1] = w0[2];
+ w0[2] = w0[3];
+ w0[3] = w1[0];
+ w1[0] = w1[1];
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 8:
- dst0[2] = src_r0[0];
- dst0[3] = src_r0[1];
- dst1[0] = src_r0[2];
- dst1[1] = src_r0[3];
- dst1[2] = src_r1[0];
- dst1[3] = src_r1[1];
- dst2[0] = src_r1[2];
- dst2[1] = src_r1[3];
break;
- case 9:
- dst0[2] = src_l0[2] | src_r0[0] << 8;
- dst0[3] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[0] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[1] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[2] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst1[3] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst2[0] = src_r1[1] >> 24 | src_r1[2] << 8;
- dst2[1] = src_r1[2] >> 24 | src_r1[3] << 8;
- dst2[2] = src_r1[3] >> 24;
- break;
+ case 1:
+ w3[2] = amd_bytealign_S ( 0, w3[0], offset_minus_4);
+ w3[1] = amd_bytealign_S (w3[0], w2[3], offset_minus_4);
+ w3[0] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
+ w2[3] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
+ w2[2] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
+ w2[1] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w2[0] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w1[3] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w1[2] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w1[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w1[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w0[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w0[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w0[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w0[0] = 0;
- case 10:
- dst0[2] = src_l0[2] | src_r0[0] << 16;
- dst0[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[0] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[1] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[2] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst1[3] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst2[0] = src_r1[1] >> 16 | src_r1[2] << 16;
- dst2[1] = src_r1[2] >> 16 | src_r1[3] << 16;
- dst2[2] = src_r1[3] >> 16;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w0[1] = w0[2];
+ w0[2] = w0[3];
+ w0[3] = w1[0];
+ w1[0] = w1[1];
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 11:
- dst0[2] = src_l0[2] | src_r0[0] << 24;
- dst0[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[0] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[1] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[2] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst1[3] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst2[0] = src_r1[1] >> 8 | src_r1[2] << 24;
- dst2[1] = src_r1[2] >> 8 | src_r1[3] << 24;
- dst2[2] = src_r1[3] >> 8;
break;
- case 12:
- dst0[3] = src_r0[0];
- dst1[0] = src_r0[1];
- dst1[1] = src_r0[2];
- dst1[2] = src_r0[3];
- dst1[3] = src_r1[0];
- dst2[0] = src_r1[1];
- dst2[1] = src_r1[2];
- dst2[2] = src_r1[3];
- break;
+ case 2:
+ w3[2] = amd_bytealign_S ( 0, w2[3], offset_minus_4);
+ w3[1] = amd_bytealign_S (w2[3], w2[2], offset_minus_4);
+ w3[0] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
+ w2[3] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
+ w2[2] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w2[1] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w2[0] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w1[3] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w1[2] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w1[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w1[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w0[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w0[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w0[1] = 0;
+ w0[0] = 0;
- case 13:
- dst0[3] = src_l0[3] | src_r0[0] << 8;
- dst1[0] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[1] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[2] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst1[3] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst2[0] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst2[1] = src_r1[1] >> 24 | src_r1[2] << 8;
- dst2[2] = src_r1[2] >> 24 | src_r1[3] << 8;
- dst2[3] = src_r1[3] >> 24;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w0[2] = w0[3];
+ w0[3] = w1[0];
+ w1[0] = w1[1];
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 14:
- dst0[3] = src_l0[3] | src_r0[0] << 16;
- dst1[0] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[1] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[2] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst1[3] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst2[0] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst2[1] = src_r1[1] >> 16 | src_r1[2] << 16;
- dst2[2] = src_r1[2] >> 16 | src_r1[3] << 16;
- dst2[3] = src_r1[3] >> 16;
break;
- case 15:
- dst0[3] = src_l0[3] | src_r0[0] << 24;
- dst1[0] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[1] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[2] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst1[3] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst2[0] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst2[1] = src_r1[1] >> 8 | src_r1[2] << 24;
- dst2[2] = src_r1[2] >> 8 | src_r1[3] << 24;
- dst2[3] = src_r1[3] >> 8;
- break;
+ case 3:
+ w3[2] = amd_bytealign_S ( 0, w2[2], offset_minus_4);
+ w3[1] = amd_bytealign_S (w2[2], w2[1], offset_minus_4);
+ w3[0] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
+ w2[3] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w2[2] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w2[1] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w2[0] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w1[3] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w1[2] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w1[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w1[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w0[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 16:
- dst1[0] = src_r0[0];
- dst1[1] = src_r0[1];
- dst1[2] = src_r0[2];
- dst1[3] = src_r0[3];
- dst2[0] = src_r1[0];
- dst2[1] = src_r1[1];
- dst2[2] = src_r1[2];
- dst2[3] = src_r1[3];
- break;
+ if (offset_mod_4 == 0)
+ {
+ w0[3] = w1[0];
+ w1[0] = w1[1];
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 17:
- dst1[0] = src_l1[0] | src_r0[0] << 8;
- dst1[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst1[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[0] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst2[1] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst2[2] = src_r1[1] >> 24 | src_r1[2] << 8;
- dst2[3] = src_r1[2] >> 24 | src_r1[3] << 8;
break;
- case 18:
- dst1[0] = src_l1[0] | src_r0[0] << 16;
- dst1[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst1[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[0] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst2[1] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst2[2] = src_r1[1] >> 16 | src_r1[2] << 16;
- dst2[3] = src_r1[2] >> 16 | src_r1[3] << 16;
- break;
+ case 4:
+ w3[2] = amd_bytealign_S ( 0, w2[1], offset_minus_4);
+ w3[1] = amd_bytealign_S (w2[1], w2[0], offset_minus_4);
+ w3[0] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w2[3] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w2[2] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w2[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w2[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w1[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w1[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w1[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w1[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 19:
- dst1[0] = src_l1[0] | src_r0[0] << 24;
- dst1[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst1[3] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[0] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst2[1] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst2[2] = src_r1[1] >> 8 | src_r1[2] << 24;
- dst2[3] = src_r1[2] >> 8 | src_r1[3] << 24;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w1[0] = w1[1];
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 20:
- dst1[1] = src_r1[0];
- dst1[2] = src_r0[1];
- dst1[3] = src_r0[2];
- dst2[0] = src_r0[3];
- dst2[1] = src_r1[0];
- dst2[2] = src_r1[1];
- dst2[3] = src_r1[2];
break;
- case 21:
- dst1[1] = src_l1[1] | src_r0[0] << 8;
- dst1[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst1[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[0] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[1] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst2[2] = src_r1[0] >> 24 | src_r1[1] << 8;
- dst2[3] = src_r1[1] >> 24 | src_r1[2] << 8;
- break;
+ case 5:
+ w3[2] = amd_bytealign_S ( 0, w2[0], offset_minus_4);
+ w3[1] = amd_bytealign_S (w2[0], w1[3], offset_minus_4);
+ w3[0] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w2[3] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w2[2] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w2[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w2[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w1[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w1[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w1[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 22:
- dst1[1] = src_l1[1] | src_r0[0] << 16;
- dst1[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst1[3] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[0] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[1] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst2[2] = src_r1[0] >> 16 | src_r1[1] << 16;
- dst2[3] = src_r1[1] >> 16 | src_r1[2] << 16;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w1[1] = w1[2];
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 23:
- dst1[1] = src_l1[1] | src_r0[0] << 24;
- dst1[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst1[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[0] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[1] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst2[2] = src_r1[0] >> 8 | src_r1[1] << 24;
- dst2[3] = src_r1[1] >> 8 | src_r1[2] << 24;
break;
- case 24:
- dst1[2] = src_r1[0];
- dst1[3] = src_r0[1];
- dst2[0] = src_r0[2];
- dst2[1] = src_r0[3];
- dst2[2] = src_r1[0];
- dst2[3] = src_r1[1];
- break;
+ case 6:
+ w3[2] = amd_bytealign_S ( 0, w1[3], offset_minus_4);
+ w3[1] = amd_bytealign_S (w1[3], w1[2], offset_minus_4);
+ w3[0] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w2[3] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w2[2] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w2[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w2[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w1[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w1[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 25:
- dst1[2] = src_l1[2] | src_r0[0] << 8;
- dst1[3] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[0] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[1] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[2] = src_r0[3] >> 24 | src_r1[0] << 8;
- dst2[3] = src_r1[0] >> 24 | src_r1[1] << 8;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w1[2] = w1[3];
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 26:
- dst1[2] = src_l1[2] | src_r0[0] << 16;
- dst1[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[0] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[1] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[2] = src_r0[3] >> 16 | src_r1[0] << 16;
- dst2[3] = src_r1[0] >> 16 | src_r1[1] << 16;
break;
- case 27:
- dst1[2] = src_l1[2] | src_r0[0] << 24;
- dst1[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[0] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[1] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[2] = src_r0[3] >> 8 | src_r1[0] << 24;
- dst2[3] = src_r1[0] >> 8 | src_r1[1] << 24;
- break;
+ case 7:
+ w3[2] = amd_bytealign_S ( 0, w1[2], offset_minus_4);
+ w3[1] = amd_bytealign_S (w1[2], w1[1], offset_minus_4);
+ w3[0] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w2[3] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w2[2] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w2[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w2[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w1[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 28:
- dst1[3] = src_r1[0];
- dst2[0] = src_r0[1];
- dst2[1] = src_r0[2];
- dst2[2] = src_r0[3];
- dst2[3] = src_r1[0];
- break;
+ if (offset_mod_4 == 0)
+ {
+ w1[3] = w2[0];
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 29:
- dst1[3] = src_l1[3] | src_r0[0] << 8;
- dst2[0] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[1] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[2] = src_r0[2] >> 24 | src_r0[3] << 8;
- dst2[3] = src_r0[3] >> 24 | src_r1[0] << 8;
break;
- case 30:
- dst1[3] = src_l1[3] | src_r0[0] << 16;
- dst2[0] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[1] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[2] = src_r0[2] >> 16 | src_r0[3] << 16;
- dst2[3] = src_r0[3] >> 16 | src_r1[0] << 16;
- break;
+ case 8:
+ w3[2] = amd_bytealign_S ( 0, w1[1], offset_minus_4);
+ w3[1] = amd_bytealign_S (w1[1], w1[0], offset_minus_4);
+ w3[0] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w2[3] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w2[2] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w2[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w2[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 31:
- dst1[3] = src_l1[3] | src_r0[0] << 24;
- dst2[0] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[1] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[2] = src_r0[2] >> 8 | src_r0[3] << 24;
- dst2[3] = src_r0[3] >> 8 | src_r1[0] << 24;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w2[0] = w2[1];
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 32:
- dst2[0] = src_r0[0];
- dst2[1] = src_r0[1];
- dst2[2] = src_r0[2];
- dst2[3] = src_r0[3];
break;
- case 33:
- dst2[0] = src_l2[0] | src_r0[0] << 8;
- dst2[1] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[2] = src_r0[1] >> 24 | src_r0[2] << 8;
- dst2[3] = src_r0[2] >> 24 | src_r0[3] << 8;
- break;
+ case 9:
+ w3[2] = amd_bytealign_S ( 0, w1[0], offset_minus_4);
+ w3[1] = amd_bytealign_S (w1[0], w0[3], offset_minus_4);
+ w3[0] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w2[3] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w2[2] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w2[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 34:
- dst2[0] = src_l2[0] | src_r0[0] << 16;
- dst2[1] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[2] = src_r0[1] >> 16 | src_r0[2] << 16;
- dst2[3] = src_r0[2] >> 16 | src_r0[3] << 16;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w2[1] = w2[2];
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 35:
- dst2[0] = src_l2[0] | src_r0[0] << 24;
- dst2[1] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[2] = src_r0[1] >> 8 | src_r0[2] << 24;
- dst2[3] = src_r0[2] >> 8 | src_r0[3] << 24;
break;
- case 36:
- dst2[1] = src_r0[0];
- dst2[2] = src_r0[1];
- dst2[3] = src_r0[2];
- break;
+ case 10:
+ w3[2] = amd_bytealign_S ( 0, w0[3], offset_minus_4);
+ w3[1] = amd_bytealign_S (w0[3], w0[2], offset_minus_4);
+ w3[0] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w2[3] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w2[2] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 37:
- dst2[1] = src_l2[1] | src_r0[0] << 8;
- dst2[2] = src_r0[0] >> 24 | src_r0[1] << 8;
- dst2[3] = src_r0[1] >> 24 | src_r0[2] << 8;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w2[2] = w2[3];
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 38:
- dst2[1] = src_l2[1] | src_r0[0] << 16;
- dst2[2] = src_r0[0] >> 16 | src_r0[1] << 16;
- dst2[3] = src_r0[1] >> 16 | src_r0[2] << 16;
break;
- case 39:
- dst2[1] = src_l2[1] | src_r0[0] << 24;
- dst2[2] = src_r0[0] >> 8 | src_r0[1] << 24;
- dst2[3] = src_r0[1] >> 8 | src_r0[2] << 24;
- break;
+ case 11:
+ w3[2] = amd_bytealign_S ( 0, w0[2], offset_minus_4);
+ w3[1] = amd_bytealign_S (w0[2], w0[1], offset_minus_4);
+ w3[0] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w2[3] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 40:
- dst2[2] = src_r0[0];
- dst2[3] = src_r0[1];
- break;
+ if (offset_mod_4 == 0)
+ {
+ w2[3] = w3[0];
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 41:
- dst2[2] = src_l2[2] | src_r0[0] << 8;
- dst2[3] = src_r0[0] >> 24 | src_r0[1] << 8;
break;
- case 42:
- dst2[2] = src_l2[2] | src_r0[0] << 16;
- dst2[3] = src_r0[0] >> 16 | src_r0[1] << 16;
- break;
+ case 12:
+ w3[2] = amd_bytealign_S ( 0, w0[1], offset_minus_4);
+ w3[1] = amd_bytealign_S (w0[1], w0[0], offset_minus_4);
+ w3[0] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 43:
- dst2[2] = src_l2[2] | src_r0[0] << 24;
- dst2[3] = src_r0[0] >> 8 | src_r0[1] << 24;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w3[0] = w3[1];
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 44:
- dst2[3] = src_r0[0];
break;
- case 45:
- dst2[3] = src_l2[3] | src_r0[0] << 8;
- break;
+ case 13:
+ w3[2] = amd_bytealign_S ( 0, w0[0], offset_minus_4);
+ w3[1] = amd_bytealign_S (w0[0], 0, offset_minus_4);
+ w3[0] = 0;
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 46:
- dst2[3] = src_l2[3] | src_r0[0] << 16;
- break;
+ if (offset_mod_4 == 0)
+ {
+ w3[1] = w3[2];
+ w3[2] = 0;
+ }
- case 47:
- dst2[3] = src_l2[3] | src_r0[0] << 24;
break;
}
-}
+ #endif
-// before: memcat16_9
-static void memcat_c15_w4x4_a3x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 append0[4], const u32 append1[4], const u32 append2[4], const u32 offset)
-{
- switch (offset)
+ #ifdef IS_NV
+ const int offset_minus_4 = 4 - (offset % 4);
+
+ const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
+
+ switch (offset / 4)
{
case 0:
- w0[0] = append0[0];
- w0[1] = append0[1];
- w0[2] = append0[2];
- w0[3] = append0[3];
- w1[0] = append1[0];
- w1[1] = append1[1];
- w1[2] = append1[2];
- w1[3] = append1[3];
- w2[0] = append2[0];
+ w3[1] = __byte_perm_S (w3[0], w3[1], selector);
+ w3[0] = __byte_perm_S (w2[3], w3[0], selector);
+ w2[3] = __byte_perm_S (w2[2], w2[3], selector);
+ w2[2] = __byte_perm_S (w2[1], w2[2], selector);
+ w2[1] = __byte_perm_S (w2[0], w2[1], selector);
+ w2[0] = __byte_perm_S (w1[3], w2[0], selector);
+ w1[3] = __byte_perm_S (w1[2], w1[3], selector);
+ w1[2] = __byte_perm_S (w1[1], w1[2], selector);
+ w1[1] = __byte_perm_S (w1[0], w1[1], selector);
+ w1[0] = __byte_perm_S (w0[3], w1[0], selector);
+ w0[3] = __byte_perm_S (w0[2], w0[3], selector);
+ w0[2] = __byte_perm_S (w0[1], w0[2], selector);
+ w0[1] = __byte_perm_S (w0[0], w0[1], selector);
+ w0[0] = __byte_perm_S ( 0, w0[0], selector);
+
break;
case 1:
- w0[0] = w0[0] | append0[0] << 8;
- w0[1] = append0[0] >> 24 | append0[1] << 8;
- w0[2] = append0[1] >> 24 | append0[2] << 8;
- w0[3] = append0[2] >> 24 | append0[3] << 8;
- w1[0] = append0[3] >> 24 | append1[0] << 8;
- w1[1] = append1[0] >> 24 | append1[1] << 8;
- w1[2] = append1[1] >> 24 | append1[2] << 8;
- w1[3] = append1[2] >> 24 | append1[3] << 8;
- w2[0] = append1[3] >> 24 | append2[0] << 8;
- w2[1] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w2[3], w3[0], selector);
+ w3[0] = __byte_perm_S (w2[2], w2[3], selector);
+ w2[3] = __byte_perm_S (w2[1], w2[2], selector);
+ w2[2] = __byte_perm_S (w2[0], w2[1], selector);
+ w2[1] = __byte_perm_S (w1[3], w2[0], selector);
+ w2[0] = __byte_perm_S (w1[2], w1[3], selector);
+ w1[3] = __byte_perm_S (w1[1], w1[2], selector);
+ w1[2] = __byte_perm_S (w1[0], w1[1], selector);
+ w1[1] = __byte_perm_S (w0[3], w1[0], selector);
+ w1[0] = __byte_perm_S (w0[2], w0[3], selector);
+ w0[3] = __byte_perm_S (w0[1], w0[2], selector);
+ w0[2] = __byte_perm_S (w0[0], w0[1], selector);
+ w0[1] = __byte_perm_S ( 0, w0[0], selector);
+ w0[0] = 0;
+
break;
case 2:
- w0[0] = w0[0] | append0[0] << 16;
- w0[1] = append0[0] >> 16 | append0[1] << 16;
- w0[2] = append0[1] >> 16 | append0[2] << 16;
- w0[3] = append0[2] >> 16 | append0[3] << 16;
- w1[0] = append0[3] >> 16 | append1[0] << 16;
- w1[1] = append1[0] >> 16 | append1[1] << 16;
- w1[2] = append1[1] >> 16 | append1[2] << 16;
- w1[3] = append1[2] >> 16 | append1[3] << 16;
- w2[0] = append1[3] >> 16 | append2[0] << 16;
- w2[1] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w2[2], w2[3], selector);
+ w3[0] = __byte_perm_S (w2[1], w2[2], selector);
+ w2[3] = __byte_perm_S (w2[0], w2[1], selector);
+ w2[2] = __byte_perm_S (w1[3], w2[0], selector);
+ w2[1] = __byte_perm_S (w1[2], w1[3], selector);
+ w2[0] = __byte_perm_S (w1[1], w1[2], selector);
+ w1[3] = __byte_perm_S (w1[0], w1[1], selector);
+ w1[2] = __byte_perm_S (w0[3], w1[0], selector);
+ w1[1] = __byte_perm_S (w0[2], w0[3], selector);
+ w1[0] = __byte_perm_S (w0[1], w0[2], selector);
+ w0[3] = __byte_perm_S (w0[0], w0[1], selector);
+ w0[2] = __byte_perm_S ( 0, w0[0], selector);
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 3:
- w0[0] = w0[0] | append0[0] << 24;
- w0[1] = append0[0] >> 8 | append0[1] << 24;
- w0[2] = append0[1] >> 8 | append0[2] << 24;
- w0[3] = append0[2] >> 8 | append0[3] << 24;
- w1[0] = append0[3] >> 8 | append1[0] << 24;
- w1[1] = append1[0] >> 8 | append1[1] << 24;
- w1[2] = append1[1] >> 8 | append1[2] << 24;
- w1[3] = append1[2] >> 8 | append1[3] << 24;
- w2[0] = append1[3] >> 8 | append2[0] << 24;
- w2[1] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w2[1], w2[2], selector);
+ w3[0] = __byte_perm_S (w2[0], w2[1], selector);
+ w2[3] = __byte_perm_S (w1[3], w2[0], selector);
+ w2[2] = __byte_perm_S (w1[2], w1[3], selector);
+ w2[1] = __byte_perm_S (w1[1], w1[2], selector);
+ w2[0] = __byte_perm_S (w1[0], w1[1], selector);
+ w1[3] = __byte_perm_S (w0[3], w1[0], selector);
+ w1[2] = __byte_perm_S (w0[2], w0[3], selector);
+ w1[1] = __byte_perm_S (w0[1], w0[2], selector);
+ w1[0] = __byte_perm_S (w0[0], w0[1], selector);
+ w0[3] = __byte_perm_S ( 0, w0[0], selector);
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 4:
- w0[1] = append0[0];
- w0[2] = append0[1];
- w0[3] = append0[2];
- w1[0] = append0[3];
- w1[1] = append1[0];
- w1[2] = append1[1];
- w1[3] = append1[2];
- w2[0] = append1[3];
- w2[1] = append2[0];
+ w3[1] = __byte_perm_S (w2[0], w2[1], selector);
+ w3[0] = __byte_perm_S (w1[3], w2[0], selector);
+ w2[3] = __byte_perm_S (w1[2], w1[3], selector);
+ w2[2] = __byte_perm_S (w1[1], w1[2], selector);
+ w2[1] = __byte_perm_S (w1[0], w1[1], selector);
+ w2[0] = __byte_perm_S (w0[3], w1[0], selector);
+ w1[3] = __byte_perm_S (w0[2], w0[3], selector);
+ w1[2] = __byte_perm_S (w0[1], w0[2], selector);
+ w1[1] = __byte_perm_S (w0[0], w0[1], selector);
+ w1[0] = __byte_perm_S ( 0, w0[0], selector);
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 5:
- w0[1] = w0[1] | append0[0] << 8;
- w0[2] = append0[0] >> 24 | append0[1] << 8;
- w0[3] = append0[1] >> 24 | append0[2] << 8;
- w1[0] = append0[2] >> 24 | append0[3] << 8;
- w1[1] = append0[3] >> 24 | append1[0] << 8;
- w1[2] = append1[0] >> 24 | append1[1] << 8;
- w1[3] = append1[1] >> 24 | append1[2] << 8;
- w2[0] = append1[2] >> 24 | append1[3] << 8;
- w2[1] = append1[3] >> 24 | append2[0] << 8;
- w2[2] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w1[3], w2[0], selector);
+ w3[0] = __byte_perm_S (w1[2], w1[3], selector);
+ w2[3] = __byte_perm_S (w1[1], w1[2], selector);
+ w2[2] = __byte_perm_S (w1[0], w1[1], selector);
+ w2[1] = __byte_perm_S (w0[3], w1[0], selector);
+ w2[0] = __byte_perm_S (w0[2], w0[3], selector);
+ w1[3] = __byte_perm_S (w0[1], w0[2], selector);
+ w1[2] = __byte_perm_S (w0[0], w0[1], selector);
+ w1[1] = __byte_perm_S ( 0, w0[0], selector);
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 6:
- w0[1] = w0[1] | append0[0] << 16;
- w0[2] = append0[0] >> 16 | append0[1] << 16;
- w0[3] = append0[1] >> 16 | append0[2] << 16;
- w1[0] = append0[2] >> 16 | append0[3] << 16;
- w1[1] = append0[3] >> 16 | append1[0] << 16;
- w1[2] = append1[0] >> 16 | append1[1] << 16;
- w1[3] = append1[1] >> 16 | append1[2] << 16;
- w2[0] = append1[2] >> 16 | append1[3] << 16;
- w2[1] = append1[3] >> 16 | append2[0] << 16;
- w2[2] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w1[2], w1[3], selector);
+ w3[0] = __byte_perm_S (w1[1], w1[2], selector);
+ w2[3] = __byte_perm_S (w1[0], w1[1], selector);
+ w2[2] = __byte_perm_S (w0[3], w1[0], selector);
+ w2[1] = __byte_perm_S (w0[2], w0[3], selector);
+ w2[0] = __byte_perm_S (w0[1], w0[2], selector);
+ w1[3] = __byte_perm_S (w0[0], w0[1], selector);
+ w1[2] = __byte_perm_S ( 0, w0[0], selector);
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 7:
- w0[1] = w0[1] | append0[0] << 24;
- w0[2] = append0[0] >> 8 | append0[1] << 24;
- w0[3] = append0[1] >> 8 | append0[2] << 24;
- w1[0] = append0[2] >> 8 | append0[3] << 24;
- w1[1] = append0[3] >> 8 | append1[0] << 24;
- w1[2] = append1[0] >> 8 | append1[1] << 24;
- w1[3] = append1[1] >> 8 | append1[2] << 24;
- w2[0] = append1[2] >> 8 | append1[3] << 24;
- w2[1] = append1[3] >> 8 | append2[0] << 24;
- w2[2] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w1[1], w1[2], selector);
+ w3[0] = __byte_perm_S (w1[0], w1[1], selector);
+ w2[3] = __byte_perm_S (w0[3], w1[0], selector);
+ w2[2] = __byte_perm_S (w0[2], w0[3], selector);
+ w2[1] = __byte_perm_S (w0[1], w0[2], selector);
+ w2[0] = __byte_perm_S (w0[0], w0[1], selector);
+ w1[3] = __byte_perm_S ( 0, w0[0], selector);
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 8:
- w0[2] = append0[0];
- w0[3] = append0[1];
- w1[0] = append0[2];
- w1[1] = append0[3];
- w1[2] = append1[0];
- w1[3] = append1[1];
- w2[0] = append1[2];
- w2[1] = append1[3];
- w2[2] = append2[0];
+ w3[1] = __byte_perm_S (w1[0], w1[1], selector);
+ w3[0] = __byte_perm_S (w0[3], w1[0], selector);
+ w2[3] = __byte_perm_S (w0[2], w0[3], selector);
+ w2[2] = __byte_perm_S (w0[1], w0[2], selector);
+ w2[1] = __byte_perm_S (w0[0], w0[1], selector);
+ w2[0] = __byte_perm_S ( 0, w0[0], selector);
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 9:
- w0[2] = w0[2] | append0[0] << 8;
- w0[3] = append0[0] >> 24 | append0[1] << 8;
- w1[0] = append0[1] >> 24 | append0[2] << 8;
- w1[1] = append0[2] >> 24 | append0[3] << 8;
- w1[2] = append0[3] >> 24 | append1[0] << 8;
- w1[3] = append1[0] >> 24 | append1[1] << 8;
- w2[0] = append1[1] >> 24 | append1[2] << 8;
- w2[1] = append1[2] >> 24 | append1[3] << 8;
- w2[2] = append1[3] >> 24 | append2[0] << 8;
- w2[3] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w0[3], w1[0], selector);
+ w3[0] = __byte_perm_S (w0[2], w0[3], selector);
+ w2[3] = __byte_perm_S (w0[1], w0[2], selector);
+ w2[2] = __byte_perm_S (w0[0], w0[1], selector);
+ w2[1] = __byte_perm_S ( 0, w0[0], selector);
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 10:
- w0[2] = w0[2] | append0[0] << 16;
- w0[3] = append0[0] >> 16 | append0[1] << 16;
- w1[0] = append0[1] >> 16 | append0[2] << 16;
- w1[1] = append0[2] >> 16 | append0[3] << 16;
- w1[2] = append0[3] >> 16 | append1[0] << 16;
- w1[3] = append1[0] >> 16 | append1[1] << 16;
- w2[0] = append1[1] >> 16 | append1[2] << 16;
- w2[1] = append1[2] >> 16 | append1[3] << 16;
- w2[2] = append1[3] >> 16 | append2[0] << 16;
- w2[3] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w0[2], w0[3], selector);
+ w3[0] = __byte_perm_S (w0[1], w0[2], selector);
+ w2[3] = __byte_perm_S (w0[0], w0[1], selector);
+ w2[2] = __byte_perm_S ( 0, w0[0], selector);
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 11:
- w0[2] = w0[2] | append0[0] << 24;
- w0[3] = append0[0] >> 8 | append0[1] << 24;
- w1[0] = append0[1] >> 8 | append0[2] << 24;
- w1[1] = append0[2] >> 8 | append0[3] << 24;
- w1[2] = append0[3] >> 8 | append1[0] << 24;
- w1[3] = append1[0] >> 8 | append1[1] << 24;
- w2[0] = append1[1] >> 8 | append1[2] << 24;
- w2[1] = append1[2] >> 8 | append1[3] << 24;
- w2[2] = append1[3] >> 8 | append2[0] << 24;
- w2[3] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w0[1], w0[2], selector);
+ w3[0] = __byte_perm_S (w0[0], w0[1], selector);
+ w2[3] = __byte_perm_S ( 0, w0[0], selector);
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
+
break;
case 12:
- w0[3] = append0[0];
- w1[0] = append0[1];
- w1[1] = append0[2];
- w1[2] = append0[3];
- w1[3] = append1[0];
- w2[0] = append1[1];
- w2[1] = append1[2];
- w2[2] = append1[3];
- w2[3] = append2[0];
- break;
+ w3[1] = __byte_perm_S (w0[0], w0[1], selector);
+ w3[0] = __byte_perm_S ( 0, w0[0], selector);
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 13:
- w0[3] = w0[3] | append0[0] << 8;
- w1[0] = append0[0] >> 24 | append0[1] << 8;
- w1[1] = append0[1] >> 24 | append0[2] << 8;
- w1[2] = append0[2] >> 24 | append0[3] << 8;
- w1[3] = append0[3] >> 24 | append1[0] << 8;
- w2[0] = append1[0] >> 24 | append1[1] << 8;
- w2[1] = append1[1] >> 24 | append1[2] << 8;
- w2[2] = append1[2] >> 24 | append1[3] << 8;
- w2[3] = append1[3] >> 24 | append2[0] << 8;
- w3[0] = append2[0] >> 24;
break;
- case 14:
- w0[3] = w0[3] | append0[0] << 16;
- w1[0] = append0[0] >> 16 | append0[1] << 16;
- w1[1] = append0[1] >> 16 | append0[2] << 16;
- w1[2] = append0[2] >> 16 | append0[3] << 16;
- w1[3] = append0[3] >> 16 | append1[0] << 16;
- w2[0] = append1[0] >> 16 | append1[1] << 16;
- w2[1] = append1[1] >> 16 | append1[2] << 16;
- w2[2] = append1[2] >> 16 | append1[3] << 16;
- w2[3] = append1[3] >> 16 | append2[0] << 16;
- w3[0] = append2[0] >> 16;
- break;
+ case 13:
+ w3[1] = __byte_perm_S ( 0, w0[0], selector);
+ w3[0] = 0;
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
- case 15:
- w0[3] = w0[3] | append0[0] << 24;
- w1[0] = append0[0] >> 8 | append0[1] << 24;
- w1[1] = append0[1] >> 8 | append0[2] << 24;
- w1[2] = append0[2] >> 8 | append0[3] << 24;
- w1[3] = append0[3] >> 8 | append1[0] << 24;
- w2[0] = append1[0] >> 8 | append1[1] << 24;
- w2[1] = append1[1] >> 8 | append1[2] << 24;
- w2[2] = append1[2] >> 8 | append1[3] << 24;
- w2[3] = append1[3] >> 8 | append2[0] << 24;
- w3[0] = append2[0] >> 8;
break;
}
+ #endif
}
-// before: memcat32_8
-static void memcat_c32_w4x4_a2x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 append0[4], const u32 append1[4], const u32 offset)
+inline void switch_buffer_by_offset_be_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 offset)
{
- switch (offset)
+ #if defined IS_AMD || defined IS_GENERIC
+ switch (offset / 4)
{
case 0:
- w0[0] = append0[0];
- w0[1] = append0[1];
- w0[2] = append0[2];
- w0[3] = append0[3];
- w1[0] = append1[0];
- w1[1] = append1[1];
- w1[2] = append1[2];
- w1[3] = append1[3];
+ w3[2] = amd_bytealign_S (w3[1], 0, offset);
+ w3[1] = amd_bytealign_S (w3[0], w3[1], offset);
+ w3[0] = amd_bytealign_S (w2[3], w3[0], offset);
+ w2[3] = amd_bytealign_S (w2[2], w2[3], offset);
+ w2[2] = amd_bytealign_S (w2[1], w2[2], offset);
+ w2[1] = amd_bytealign_S (w2[0], w2[1], offset);
+ w2[0] = amd_bytealign_S (w1[3], w2[0], offset);
+ w1[3] = amd_bytealign_S (w1[2], w1[3], offset);
+ w1[2] = amd_bytealign_S (w1[1], w1[2], offset);
+ w1[1] = amd_bytealign_S (w1[0], w1[1], offset);
+ w1[0] = amd_bytealign_S (w0[3], w1[0], offset);
+ w0[3] = amd_bytealign_S (w0[2], w0[3], offset);
+ w0[2] = amd_bytealign_S (w0[1], w0[2], offset);
+ w0[1] = amd_bytealign_S (w0[0], w0[1], offset);
+ w0[0] = amd_bytealign_S ( 0, w0[0], offset);
break;
case 1:
- w0[0] = w0[0] | append0[0] << 8;
- w0[1] = append0[0] >> 24 | append0[1] << 8;
- w0[2] = append0[1] >> 24 | append0[2] << 8;
- w0[3] = append0[2] >> 24 | append0[3] << 8;
- w1[0] = append0[3] >> 24 | append1[0] << 8;
- w1[1] = append1[0] >> 24 | append1[1] << 8;
- w1[2] = append1[1] >> 24 | append1[2] << 8;
- w1[3] = append1[2] >> 24 | append1[3] << 8;
- w2[0] = append1[3] >> 24;
+ w3[2] = amd_bytealign_S (w3[0], 0, offset);
+ w3[1] = amd_bytealign_S (w2[3], w3[0], offset);
+ w3[0] = amd_bytealign_S (w2[2], w2[3], offset);
+ w2[3] = amd_bytealign_S (w2[1], w2[2], offset);
+ w2[2] = amd_bytealign_S (w2[0], w2[1], offset);
+ w2[1] = amd_bytealign_S (w1[3], w2[0], offset);
+ w2[0] = amd_bytealign_S (w1[2], w1[3], offset);
+ w1[3] = amd_bytealign_S (w1[1], w1[2], offset);
+ w1[2] = amd_bytealign_S (w1[0], w1[1], offset);
+ w1[1] = amd_bytealign_S (w0[3], w1[0], offset);
+ w1[0] = amd_bytealign_S (w0[2], w0[3], offset);
+ w0[3] = amd_bytealign_S (w0[1], w0[2], offset);
+ w0[2] = amd_bytealign_S (w0[0], w0[1], offset);
+ w0[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w0[0] = 0;
break;
case 2:
- w0[0] = w0[0] | append0[0] << 16;
- w0[1] = append0[0] >> 16 | append0[1] << 16;
- w0[2] = append0[1] >> 16 | append0[2] << 16;
- w0[3] = append0[2] >> 16 | append0[3] << 16;
- w1[0] = append0[3] >> 16 | append1[0] << 16;
- w1[1] = append1[0] >> 16 | append1[1] << 16;
- w1[2] = append1[1] >> 16 | append1[2] << 16;
- w1[3] = append1[2] >> 16 | append1[3] << 16;
- w2[0] = append1[3] >> 16;
+ w3[2] = amd_bytealign_S (w2[3], 0, offset);
+ w3[1] = amd_bytealign_S (w2[2], w2[3], offset);
+ w3[0] = amd_bytealign_S (w2[1], w2[2], offset);
+ w2[3] = amd_bytealign_S (w2[0], w2[1], offset);
+ w2[2] = amd_bytealign_S (w1[3], w2[0], offset);
+ w2[1] = amd_bytealign_S (w1[2], w1[3], offset);
+ w2[0] = amd_bytealign_S (w1[1], w1[2], offset);
+ w1[3] = amd_bytealign_S (w1[0], w1[1], offset);
+ w1[2] = amd_bytealign_S (w0[3], w1[0], offset);
+ w1[1] = amd_bytealign_S (w0[2], w0[3], offset);
+ w1[0] = amd_bytealign_S (w0[1], w0[2], offset);
+ w0[3] = amd_bytealign_S (w0[0], w0[1], offset);
+ w0[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 3:
- w0[0] = w0[0] | append0[0] << 24;
- w0[1] = append0[0] >> 8 | append0[1] << 24;
- w0[2] = append0[1] >> 8 | append0[2] << 24;
- w0[3] = append0[2] >> 8 | append0[3] << 24;
- w1[0] = append0[3] >> 8 | append1[0] << 24;
- w1[1] = append1[0] >> 8 | append1[1] << 24;
- w1[2] = append1[1] >> 8 | append1[2] << 24;
- w1[3] = append1[2] >> 8 | append1[3] << 24;
- w2[0] = append1[3] >> 8;
+ w3[2] = amd_bytealign_S (w2[2], 0, offset);
+ w3[1] = amd_bytealign_S (w2[1], w2[2], offset);
+ w3[0] = amd_bytealign_S (w2[0], w2[1], offset);
+ w2[3] = amd_bytealign_S (w1[3], w2[0], offset);
+ w2[2] = amd_bytealign_S (w1[2], w1[3], offset);
+ w2[1] = amd_bytealign_S (w1[1], w1[2], offset);
+ w2[0] = amd_bytealign_S (w1[0], w1[1], offset);
+ w1[3] = amd_bytealign_S (w0[3], w1[0], offset);
+ w1[2] = amd_bytealign_S (w0[2], w0[3], offset);
+ w1[1] = amd_bytealign_S (w0[1], w0[2], offset);
+ w1[0] = amd_bytealign_S (w0[0], w0[1], offset);
+ w0[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 4:
- w0[1] = append0[0];
- w0[2] = append0[1];
- w0[3] = append0[2];
- w1[0] = append0[3];
- w1[1] = append1[0];
- w1[2] = append1[1];
- w1[3] = append1[2];
- w2[0] = append1[3];
+ w3[2] = amd_bytealign_S (w2[1], 0, offset);
+ w3[1] = amd_bytealign_S (w2[0], w2[1], offset);
+ w3[0] = amd_bytealign_S (w1[3], w2[0], offset);
+ w2[3] = amd_bytealign_S (w1[2], w1[3], offset);
+ w2[2] = amd_bytealign_S (w1[1], w1[2], offset);
+ w2[1] = amd_bytealign_S (w1[0], w1[1], offset);
+ w2[0] = amd_bytealign_S (w0[3], w1[0], offset);
+ w1[3] = amd_bytealign_S (w0[2], w0[3], offset);
+ w1[2] = amd_bytealign_S (w0[1], w0[2], offset);
+ w1[1] = amd_bytealign_S (w0[0], w0[1], offset);
+ w1[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 5:
- w0[1] = w0[1] | append0[0] << 8;
- w0[2] = append0[0] >> 24 | append0[1] << 8;
- w0[3] = append0[1] >> 24 | append0[2] << 8;
- w1[0] = append0[2] >> 24 | append0[3] << 8;
- w1[1] = append0[3] >> 24 | append1[0] << 8;
- w1[2] = append1[0] >> 24 | append1[1] << 8;
- w1[3] = append1[1] >> 24 | append1[2] << 8;
- w2[0] = append1[2] >> 24 | append1[3] << 8;
- w2[1] = append1[3] >> 24;
+ w3[2] = amd_bytealign_S (w2[0], 0, offset);
+ w3[1] = amd_bytealign_S (w1[3], w2[0], offset);
+ w3[0] = amd_bytealign_S (w1[2], w1[3], offset);
+ w2[3] = amd_bytealign_S (w1[1], w1[2], offset);
+ w2[2] = amd_bytealign_S (w1[0], w1[1], offset);
+ w2[1] = amd_bytealign_S (w0[3], w1[0], offset);
+ w2[0] = amd_bytealign_S (w0[2], w0[3], offset);
+ w1[3] = amd_bytealign_S (w0[1], w0[2], offset);
+ w1[2] = amd_bytealign_S (w0[0], w0[1], offset);
+ w1[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 6:
- w0[1] = w0[1] | append0[0] << 16;
- w0[2] = append0[0] >> 16 | append0[1] << 16;
- w0[3] = append0[1] >> 16 | append0[2] << 16;
- w1[0] = append0[2] >> 16 | append0[3] << 16;
- w1[1] = append0[3] >> 16 | append1[0] << 16;
- w1[2] = append1[0] >> 16 | append1[1] << 16;
- w1[3] = append1[1] >> 16 | append1[2] << 16;
- w2[0] = append1[2] >> 16 | append1[3] << 16;
- w2[1] = append1[3] >> 16;
+ w3[2] = amd_bytealign_S (w1[3], 0, offset);
+ w3[1] = amd_bytealign_S (w1[2], w1[3], offset);
+ w3[0] = amd_bytealign_S (w1[1], w1[2], offset);
+ w2[3] = amd_bytealign_S (w1[0], w1[1], offset);
+ w2[2] = amd_bytealign_S (w0[3], w1[0], offset);
+ w2[1] = amd_bytealign_S (w0[2], w0[3], offset);
+ w2[0] = amd_bytealign_S (w0[1], w0[2], offset);
+ w1[3] = amd_bytealign_S (w0[0], w0[1], offset);
+ w1[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 7:
- w0[1] = w0[1] | append0[0] << 24;
- w0[2] = append0[0] >> 8 | append0[1] << 24;
- w0[3] = append0[1] >> 8 | append0[2] << 24;
- w1[0] = append0[2] >> 8 | append0[3] << 24;
- w1[1] = append0[3] >> 8 | append1[0] << 24;
- w1[2] = append1[0] >> 8 | append1[1] << 24;
- w1[3] = append1[1] >> 8 | append1[2] << 24;
- w2[0] = append1[2] >> 8 | append1[3] << 24;
- w2[1] = append1[3] >> 8;
+ w3[2] = amd_bytealign_S (w1[2], 0, offset);
+ w3[1] = amd_bytealign_S (w1[1], w1[2], offset);
+ w3[0] = amd_bytealign_S (w1[0], w1[1], offset);
+ w2[3] = amd_bytealign_S (w0[3], w1[0], offset);
+ w2[2] = amd_bytealign_S (w0[2], w0[3], offset);
+ w2[1] = amd_bytealign_S (w0[1], w0[2], offset);
+ w2[0] = amd_bytealign_S (w0[0], w0[1], offset);
+ w1[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 8:
- w0[2] = append0[0];
- w0[3] = append0[1];
- w1[0] = append0[2];
- w1[1] = append0[3];
- w1[2] = append1[0];
- w1[3] = append1[1];
- w2[0] = append1[2];
- w2[1] = append1[3];
+ w3[2] = amd_bytealign_S (w1[1], 0, offset);
+ w3[1] = amd_bytealign_S (w1[0], w1[1], offset);
+ w3[0] = amd_bytealign_S (w0[3], w1[0], offset);
+ w2[3] = amd_bytealign_S (w0[2], w0[3], offset);
+ w2[2] = amd_bytealign_S (w0[1], w0[2], offset);
+ w2[1] = amd_bytealign_S (w0[0], w0[1], offset);
+ w2[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 9:
- w0[2] = w0[2] | append0[0] << 8;
- w0[3] = append0[0] >> 24 | append0[1] << 8;
- w1[0] = append0[1] >> 24 | append0[2] << 8;
- w1[1] = append0[2] >> 24 | append0[3] << 8;
- w1[2] = append0[3] >> 24 | append1[0] << 8;
- w1[3] = append1[0] >> 24 | append1[1] << 8;
- w2[0] = append1[1] >> 24 | append1[2] << 8;
- w2[1] = append1[2] >> 24 | append1[3] << 8;
- w2[2] = append1[3] >> 24;
+ w3[2] = amd_bytealign_S (w1[0], 0, offset);
+ w3[1] = amd_bytealign_S (w0[3], w1[0], offset);
+ w3[0] = amd_bytealign_S (w0[2], w0[3], offset);
+ w2[3] = amd_bytealign_S (w0[1], w0[2], offset);
+ w2[2] = amd_bytealign_S (w0[0], w0[1], offset);
+ w2[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 10:
- w0[2] = w0[2] | append0[0] << 16;
- w0[3] = append0[0] >> 16 | append0[1] << 16;
- w1[0] = append0[1] >> 16 | append0[2] << 16;
- w1[1] = append0[2] >> 16 | append0[3] << 16;
- w1[2] = append0[3] >> 16 | append1[0] << 16;
- w1[3] = append1[0] >> 16 | append1[1] << 16;
- w2[0] = append1[1] >> 16 | append1[2] << 16;
- w2[1] = append1[2] >> 16 | append1[3] << 16;
- w2[2] = append1[3] >> 16;
+ w3[2] = amd_bytealign_S (w0[3], 0, offset);
+ w3[1] = amd_bytealign_S (w0[2], w0[3], offset);
+ w3[0] = amd_bytealign_S (w0[1], w0[2], offset);
+ w2[3] = amd_bytealign_S (w0[0], w0[1], offset);
+ w2[2] = amd_bytealign_S ( 0, w0[0], offset);
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 11:
- w0[2] = w0[2] | append0[0] << 24;
- w0[3] = append0[0] >> 8 | append0[1] << 24;
- w1[0] = append0[1] >> 8 | append0[2] << 24;
- w1[1] = append0[2] >> 8 | append0[3] << 24;
- w1[2] = append0[3] >> 8 | append1[0] << 24;
- w1[3] = append1[0] >> 8 | append1[1] << 24;
- w2[0] = append1[1] >> 8 | append1[2] << 24;
- w2[1] = append1[2] >> 8 | append1[3] << 24;
- w2[2] = append1[3] >> 8;
+ w3[2] = amd_bytealign_S (w0[2], 0, offset);
+ w3[1] = amd_bytealign_S (w0[1], w0[2], offset);
+ w3[0] = amd_bytealign_S (w0[0], w0[1], offset);
+ w2[3] = amd_bytealign_S ( 0, w0[0], offset);
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 12:
- w0[3] = append0[0];
- w1[0] = append0[1];
- w1[1] = append0[2];
- w1[2] = append0[3];
- w1[3] = append1[0];
- w2[0] = append1[1];
- w2[1] = append1[2];
- w2[2] = append1[3];
+ w3[2] = amd_bytealign_S (w0[1], 0, offset);
+ w3[1] = amd_bytealign_S (w0[0], w0[1], offset);
+ w3[0] = amd_bytealign_S ( 0, w0[0], offset);
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 13:
- w0[3] = w0[3] | append0[0] << 8;
- w1[0] = append0[0] >> 24 | append0[1] << 8;
- w1[1] = append0[1] >> 24 | append0[2] << 8;
- w1[2] = append0[2] >> 24 | append0[3] << 8;
- w1[3] = append0[3] >> 24 | append1[0] << 8;
- w2[0] = append1[0] >> 24 | append1[1] << 8;
- w2[1] = append1[1] >> 24 | append1[2] << 8;
- w2[2] = append1[2] >> 24 | append1[3] << 8;
- w2[3] = append1[3] >> 24;
- break;
-
- case 14:
- w0[3] = w0[3] | append0[0] << 16;
- w1[0] = append0[0] >> 16 | append0[1] << 16;
- w1[1] = append0[1] >> 16 | append0[2] << 16;
- w1[2] = append0[2] >> 16 | append0[3] << 16;
- w1[3] = append0[3] >> 16 | append1[0] << 16;
- w2[0] = append1[0] >> 16 | append1[1] << 16;
- w2[1] = append1[1] >> 16 | append1[2] << 16;
- w2[2] = append1[2] >> 16 | append1[3] << 16;
- w2[3] = append1[3] >> 16;
- break;
-
- case 15:
- w0[3] = w0[3] | append0[0] << 24;
- w1[0] = append0[0] >> 8 | append0[1] << 24;
- w1[1] = append0[1] >> 8 | append0[2] << 24;
- w1[2] = append0[2] >> 8 | append0[3] << 24;
- w1[3] = append0[3] >> 8 | append1[0] << 24;
- w2[0] = append1[0] >> 8 | append1[1] << 24;
- w2[1] = append1[1] >> 8 | append1[2] << 24;
- w2[2] = append1[2] >> 8 | append1[3] << 24;
- w2[3] = append1[3] >> 8;
- break;
-
- case 16:
- w1[0] = append0[0];
- w1[1] = append0[1];
- w1[2] = append0[2];
- w1[3] = append0[3];
- w2[0] = append1[0];
- w2[1] = append1[1];
- w2[2] = append1[2];
- w2[3] = append1[3];
- break;
-
- case 17:
- w1[0] = w1[0] | append0[0] << 8;
- w1[1] = append0[0] >> 24 | append0[1] << 8;
- w1[2] = append0[1] >> 24 | append0[2] << 8;
- w1[3] = append0[2] >> 24 | append0[3] << 8;
- w2[0] = append0[3] >> 24 | append1[0] << 8;
- w2[1] = append1[0] >> 24 | append1[1] << 8;
- w2[2] = append1[1] >> 24 | append1[2] << 8;
- w2[3] = append1[2] >> 24 | append1[3] << 8;
- w3[0] = append1[3] >> 24;
- break;
-
- case 18:
- w1[0] = w1[0] | append0[0] << 16;
- w1[1] = append0[0] >> 16 | append0[1] << 16;
- w1[2] = append0[1] >> 16 | append0[2] << 16;
- w1[3] = append0[2] >> 16 | append0[3] << 16;
- w2[0] = append0[3] >> 16 | append1[0] << 16;
- w2[1] = append1[0] >> 16 | append1[1] << 16;
- w2[2] = append1[1] >> 16 | append1[2] << 16;
- w2[3] = append1[2] >> 16 | append1[3] << 16;
- w3[0] = append1[3] >> 16;
- break;
-
- case 19:
- w1[0] = w1[0] | append0[0] << 24;
- w1[1] = append0[0] >> 8 | append0[1] << 24;
- w1[2] = append0[1] >> 8 | append0[2] << 24;
- w1[3] = append0[2] >> 8 | append0[3] << 24;
- w2[0] = append0[3] >> 8 | append1[0] << 24;
- w2[1] = append1[0] >> 8 | append1[1] << 24;
- w2[2] = append1[1] >> 8 | append1[2] << 24;
- w2[3] = append1[2] >> 8 | append1[3] << 24;
- w3[0] = append1[3] >> 8;
- break;
-
- case 20:
- w1[1] = append0[0];
- w1[2] = append0[1];
- w1[3] = append0[2];
- w2[0] = append0[3];
- w2[1] = append1[0];
- w2[2] = append1[1];
- w2[3] = append1[2];
- w3[0] = append1[3];
- break;
-
- case 21:
- w1[1] = w1[1] | append0[0] << 8;
- w1[2] = append0[0] >> 24 | append0[1] << 8;
- w1[3] = append0[1] >> 24 | append0[2] << 8;
- w2[0] = append0[2] >> 24 | append0[3] << 8;
- w2[1] = append0[3] >> 24 | append1[0] << 8;
- w2[2] = append1[0] >> 24 | append1[1] << 8;
- w2[3] = append1[1] >> 24 | append1[2] << 8;
- w3[0] = append1[2] >> 24 | append1[3] << 8;
- w3[1] = append1[3] >> 24;
- break;
-
- case 22:
- w1[1] = w1[1] | append0[0] << 16;
- w1[2] = append0[0] >> 16 | append0[1] << 16;
- w1[3] = append0[1] >> 16 | append0[2] << 16;
- w2[0] = append0[2] >> 16 | append0[3] << 16;
- w2[1] = append0[3] >> 16 | append1[0] << 16;
- w2[2] = append1[0] >> 16 | append1[1] << 16;
- w2[3] = append1[1] >> 16 | append1[2] << 16;
- w3[0] = append1[2] >> 16 | append1[3] << 16;
- w3[1] = append1[3] >> 16;
- break;
-
- case 23:
- w1[1] = w1[1] | append0[0] << 24;
- w1[2] = append0[0] >> 8 | append0[1] << 24;
- w1[3] = append0[1] >> 8 | append0[2] << 24;
- w2[0] = append0[2] >> 8 | append0[3] << 24;
- w2[1] = append0[3] >> 8 | append1[0] << 24;
- w2[2] = append1[0] >> 8 | append1[1] << 24;
- w2[3] = append1[1] >> 8 | append1[2] << 24;
- w3[0] = append1[2] >> 8 | append1[3] << 24;
- w3[1] = append1[3] >> 8;
- break;
-
- case 24:
- w1[2] = append0[0];
- w1[3] = append0[1];
- w2[0] = append0[2];
- w2[1] = append0[3];
- w2[2] = append1[0];
- w2[3] = append1[1];
- w3[0] = append1[2];
- w3[1] = append1[3];
- break;
-
- case 25:
- w1[2] = w1[2] | append0[0] << 8;
- w1[3] = append0[0] >> 24 | append0[1] << 8;
- w2[0] = append0[1] >> 24 | append0[2] << 8;
- w2[1] = append0[2] >> 24 | append0[3] << 8;
- w2[2] = append0[3] >> 24 | append1[0] << 8;
- w2[3] = append1[0] >> 24 | append1[1] << 8;
- w3[0] = append1[1] >> 24 | append1[2] << 8;
- w3[1] = append1[2] >> 24 | append1[3] << 8;
- break;
-
- case 26:
- w1[2] = w1[2] | append0[0] << 16;
- w1[3] = append0[0] >> 16 | append0[1] << 16;
- w2[0] = append0[1] >> 16 | append0[2] << 16;
- w2[1] = append0[2] >> 16 | append0[3] << 16;
- w2[2] = append0[3] >> 16 | append1[0] << 16;
- w2[3] = append1[0] >> 16 | append1[1] << 16;
- w3[0] = append1[1] >> 16 | append1[2] << 16;
- w3[1] = append1[2] >> 16 | append1[3] << 16;
- break;
-
- case 27:
- w1[2] = w1[2] | append0[0] << 24;
- w1[3] = append0[0] >> 8 | append0[1] << 24;
- w2[0] = append0[1] >> 8 | append0[2] << 24;
- w2[1] = append0[2] >> 8 | append0[3] << 24;
- w2[2] = append0[3] >> 8 | append1[0] << 24;
- w2[3] = append1[0] >> 8 | append1[1] << 24;
- w3[0] = append1[1] >> 8 | append1[2] << 24;
- w3[1] = append1[2] >> 8 | append1[3] << 24;
- break;
-
- case 28:
- w1[3] = append0[0];
- w2[0] = append0[1];
- w2[1] = append0[2];
- w2[2] = append0[3];
- w2[3] = append1[0];
- w3[0] = append1[1];
- w3[1] = append1[2];
- break;
-
- case 29:
- w1[3] = w1[3] | append0[0] << 8;
- w2[0] = append0[0] >> 24 | append0[1] << 8;
- w2[1] = append0[1] >> 24 | append0[2] << 8;
- w2[2] = append0[2] >> 24 | append0[3] << 8;
- w2[3] = append0[3] >> 24 | append1[0] << 8;
- w3[0] = append1[0] >> 24 | append1[1] << 8;
- w3[1] = append1[1] >> 24 | append1[2] << 8;
- break;
-
- case 30:
- w1[3] = w1[3] | append0[0] << 16;
- w2[0] = append0[0] >> 16 | append0[1] << 16;
- w2[1] = append0[1] >> 16 | append0[2] << 16;
- w2[2] = append0[2] >> 16 | append0[3] << 16;
- w2[3] = append0[3] >> 16 | append1[0] << 16;
- w3[0] = append1[0] >> 16 | append1[1] << 16;
- w3[1] = append1[1] >> 16 | append1[2] << 16;
- break;
-
- case 31:
- w1[3] = w1[3] | append0[0] << 24;
- w2[0] = append0[0] >> 8 | append0[1] << 24;
- w2[1] = append0[1] >> 8 | append0[2] << 24;
- w2[2] = append0[2] >> 8 | append0[3] << 24;
- w2[3] = append0[3] >> 8 | append1[0] << 24;
- w3[0] = append1[0] >> 8 | append1[1] << 24;
- w3[1] = append1[1] >> 8 | append1[2] << 24;
- break;
-
- case 32:
- w2[0] = append0[0];
- w2[1] = append0[1];
- w2[2] = append0[2];
- w2[3] = append0[3];
- w3[0] = append1[0];
- w3[1] = append1[1];
+ w3[2] = amd_bytealign_S (w0[0], 0, offset);
+ w3[1] = amd_bytealign_S ( 0, w0[0], offset);
+ w3[0] = 0;
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
}
-}
+ #endif
-// before: memcat32_9
-static void memcat_c32_w4x4_a3x4 (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 append0[4], const u32 append1[4], const u32 append2[4], const u32 offset)
-{
- switch (offset)
+ #ifdef IS_NV
+ const int selector = (0x76543210 >> ((offset & 3) * 4)) & 0xffff;
+
+ switch (offset / 4)
{
case 0:
- w0[0] = append0[0];
- w0[1] = append0[1];
- w0[2] = append0[2];
- w0[3] = append0[3];
- w1[0] = append1[0];
- w1[1] = append1[1];
- w1[2] = append1[2];
- w1[3] = append1[3];
- w2[0] = append2[0];
+ w3[1] = __byte_perm_S (w3[1], w3[0], selector);
+ w3[0] = __byte_perm_S (w3[0], w2[3], selector);
+ w2[3] = __byte_perm_S (w2[3], w2[2], selector);
+ w2[2] = __byte_perm_S (w2[2], w2[1], selector);
+ w2[1] = __byte_perm_S (w2[1], w2[0], selector);
+ w2[0] = __byte_perm_S (w2[0], w1[3], selector);
+ w1[3] = __byte_perm_S (w1[3], w1[2], selector);
+ w1[2] = __byte_perm_S (w1[2], w1[1], selector);
+ w1[1] = __byte_perm_S (w1[1], w1[0], selector);
+ w1[0] = __byte_perm_S (w1[0], w0[3], selector);
+ w0[3] = __byte_perm_S (w0[3], w0[2], selector);
+ w0[2] = __byte_perm_S (w0[2], w0[1], selector);
+ w0[1] = __byte_perm_S (w0[1], w0[0], selector);
+ w0[0] = __byte_perm_S (w0[0], 0, selector);
break;
case 1:
- w0[0] = w0[0] | append0[0] << 8;
- w0[1] = append0[0] >> 24 | append0[1] << 8;
- w0[2] = append0[1] >> 24 | append0[2] << 8;
- w0[3] = append0[2] >> 24 | append0[3] << 8;
- w1[0] = append0[3] >> 24 | append1[0] << 8;
- w1[1] = append1[0] >> 24 | append1[1] << 8;
- w1[2] = append1[1] >> 24 | append1[2] << 8;
- w1[3] = append1[2] >> 24 | append1[3] << 8;
- w2[0] = append1[3] >> 24 | append2[0] << 8;
- w2[1] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w3[0], w2[3], selector);
+ w3[0] = __byte_perm_S (w2[3], w2[2], selector);
+ w2[3] = __byte_perm_S (w2[2], w2[1], selector);
+ w2[2] = __byte_perm_S (w2[1], w2[0], selector);
+ w2[1] = __byte_perm_S (w2[0], w1[3], selector);
+ w2[0] = __byte_perm_S (w1[3], w1[2], selector);
+ w1[3] = __byte_perm_S (w1[2], w1[1], selector);
+ w1[2] = __byte_perm_S (w1[1], w1[0], selector);
+ w1[1] = __byte_perm_S (w1[0], w0[3], selector);
+ w1[0] = __byte_perm_S (w0[3], w0[2], selector);
+ w0[3] = __byte_perm_S (w0[2], w0[1], selector);
+ w0[2] = __byte_perm_S (w0[1], w0[0], selector);
+ w0[1] = __byte_perm_S (w0[0], 0, selector);
+ w0[0] = 0;
break;
case 2:
- w0[0] = w0[0] | append0[0] << 16;
- w0[1] = append0[0] >> 16 | append0[1] << 16;
- w0[2] = append0[1] >> 16 | append0[2] << 16;
- w0[3] = append0[2] >> 16 | append0[3] << 16;
- w1[0] = append0[3] >> 16 | append1[0] << 16;
- w1[1] = append1[0] >> 16 | append1[1] << 16;
- w1[2] = append1[1] >> 16 | append1[2] << 16;
- w1[3] = append1[2] >> 16 | append1[3] << 16;
- w2[0] = append1[3] >> 16 | append2[0] << 16;
- w2[1] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w2[3], w2[2], selector);
+ w3[0] = __byte_perm_S (w2[2], w2[1], selector);
+ w2[3] = __byte_perm_S (w2[1], w2[0], selector);
+ w2[2] = __byte_perm_S (w2[0], w1[3], selector);
+ w2[1] = __byte_perm_S (w1[3], w1[2], selector);
+ w2[0] = __byte_perm_S (w1[2], w1[1], selector);
+ w1[3] = __byte_perm_S (w1[1], w1[0], selector);
+ w1[2] = __byte_perm_S (w1[0], w0[3], selector);
+ w1[1] = __byte_perm_S (w0[3], w0[2], selector);
+ w1[0] = __byte_perm_S (w0[2], w0[1], selector);
+ w0[3] = __byte_perm_S (w0[1], w0[0], selector);
+ w0[2] = __byte_perm_S (w0[0], 0, selector);
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 3:
- w0[0] = w0[0] | append0[0] << 24;
- w0[1] = append0[0] >> 8 | append0[1] << 24;
- w0[2] = append0[1] >> 8 | append0[2] << 24;
- w0[3] = append0[2] >> 8 | append0[3] << 24;
- w1[0] = append0[3] >> 8 | append1[0] << 24;
- w1[1] = append1[0] >> 8 | append1[1] << 24;
- w1[2] = append1[1] >> 8 | append1[2] << 24;
- w1[3] = append1[2] >> 8 | append1[3] << 24;
- w2[0] = append1[3] >> 8 | append2[0] << 24;
- w2[1] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w2[2], w2[1], selector);
+ w3[0] = __byte_perm_S (w2[1], w2[0], selector);
+ w2[3] = __byte_perm_S (w2[0], w1[3], selector);
+ w2[2] = __byte_perm_S (w1[3], w1[2], selector);
+ w2[1] = __byte_perm_S (w1[2], w1[1], selector);
+ w2[0] = __byte_perm_S (w1[1], w1[0], selector);
+ w1[3] = __byte_perm_S (w1[0], w0[3], selector);
+ w1[2] = __byte_perm_S (w0[3], w0[2], selector);
+ w1[1] = __byte_perm_S (w0[2], w0[1], selector);
+ w1[0] = __byte_perm_S (w0[1], w0[0], selector);
+ w0[3] = __byte_perm_S (w0[0], 0, selector);
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 4:
- w0[1] = append0[0];
- w0[2] = append0[1];
- w0[3] = append0[2];
- w1[0] = append0[3];
- w1[1] = append1[0];
- w1[2] = append1[1];
- w1[3] = append1[2];
- w2[0] = append1[3];
- w2[1] = append2[0];
+ w3[1] = __byte_perm_S (w2[1], w2[0], selector);
+ w3[0] = __byte_perm_S (w2[0], w1[3], selector);
+ w2[3] = __byte_perm_S (w1[3], w1[2], selector);
+ w2[2] = __byte_perm_S (w1[2], w1[1], selector);
+ w2[1] = __byte_perm_S (w1[1], w1[0], selector);
+ w2[0] = __byte_perm_S (w1[0], w0[3], selector);
+ w1[3] = __byte_perm_S (w0[3], w0[2], selector);
+ w1[2] = __byte_perm_S (w0[2], w0[1], selector);
+ w1[1] = __byte_perm_S (w0[1], w0[0], selector);
+ w1[0] = __byte_perm_S (w0[0], 0, selector);
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 5:
- w0[1] = w0[1] | append0[0] << 8;
- w0[2] = append0[0] >> 24 | append0[1] << 8;
- w0[3] = append0[1] >> 24 | append0[2] << 8;
- w1[0] = append0[2] >> 24 | append0[3] << 8;
- w1[1] = append0[3] >> 24 | append1[0] << 8;
- w1[2] = append1[0] >> 24 | append1[1] << 8;
- w1[3] = append1[1] >> 24 | append1[2] << 8;
- w2[0] = append1[2] >> 24 | append1[3] << 8;
- w2[1] = append1[3] >> 24 | append2[0] << 8;
- w2[2] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w2[0], w1[3], selector);
+ w3[0] = __byte_perm_S (w1[3], w1[2], selector);
+ w2[3] = __byte_perm_S (w1[2], w1[1], selector);
+ w2[2] = __byte_perm_S (w1[1], w1[0], selector);
+ w2[1] = __byte_perm_S (w1[0], w0[3], selector);
+ w2[0] = __byte_perm_S (w0[3], w0[2], selector);
+ w1[3] = __byte_perm_S (w0[2], w0[1], selector);
+ w1[2] = __byte_perm_S (w0[1], w0[0], selector);
+ w1[1] = __byte_perm_S (w0[0], 0, selector);
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 6:
- w0[1] = w0[1] | append0[0] << 16;
- w0[2] = append0[0] >> 16 | append0[1] << 16;
- w0[3] = append0[1] >> 16 | append0[2] << 16;
- w1[0] = append0[2] >> 16 | append0[3] << 16;
- w1[1] = append0[3] >> 16 | append1[0] << 16;
- w1[2] = append1[0] >> 16 | append1[1] << 16;
- w1[3] = append1[1] >> 16 | append1[2] << 16;
- w2[0] = append1[2] >> 16 | append1[3] << 16;
- w2[1] = append1[3] >> 16 | append2[0] << 16;
- w2[2] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w1[3], w1[2], selector);
+ w3[0] = __byte_perm_S (w1[2], w1[1], selector);
+ w2[3] = __byte_perm_S (w1[1], w1[0], selector);
+ w2[2] = __byte_perm_S (w1[0], w0[3], selector);
+ w2[1] = __byte_perm_S (w0[3], w0[2], selector);
+ w2[0] = __byte_perm_S (w0[2], w0[1], selector);
+ w1[3] = __byte_perm_S (w0[1], w0[0], selector);
+ w1[2] = __byte_perm_S (w0[0], 0, selector);
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 7:
- w0[1] = w0[1] | append0[0] << 24;
- w0[2] = append0[0] >> 8 | append0[1] << 24;
- w0[3] = append0[1] >> 8 | append0[2] << 24;
- w1[0] = append0[2] >> 8 | append0[3] << 24;
- w1[1] = append0[3] >> 8 | append1[0] << 24;
- w1[2] = append1[0] >> 8 | append1[1] << 24;
- w1[3] = append1[1] >> 8 | append1[2] << 24;
- w2[0] = append1[2] >> 8 | append1[3] << 24;
- w2[1] = append1[3] >> 8 | append2[0] << 24;
- w2[2] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w1[2], w1[1], selector);
+ w3[0] = __byte_perm_S (w1[1], w1[0], selector);
+ w2[3] = __byte_perm_S (w1[0], w0[3], selector);
+ w2[2] = __byte_perm_S (w0[3], w0[2], selector);
+ w2[1] = __byte_perm_S (w0[2], w0[1], selector);
+ w2[0] = __byte_perm_S (w0[1], w0[0], selector);
+ w1[3] = __byte_perm_S (w0[0], 0, selector);
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 8:
- w0[2] = append0[0];
- w0[3] = append0[1];
- w1[0] = append0[2];
- w1[1] = append0[3];
- w1[2] = append1[0];
- w1[3] = append1[1];
- w2[0] = append1[2];
- w2[1] = append1[3];
- w2[2] = append2[0];
+ w3[1] = __byte_perm_S (w1[1], w1[0], selector);
+ w3[0] = __byte_perm_S (w1[0], w0[3], selector);
+ w2[3] = __byte_perm_S (w0[3], w0[2], selector);
+ w2[2] = __byte_perm_S (w0[2], w0[1], selector);
+ w2[1] = __byte_perm_S (w0[1], w0[0], selector);
+ w2[0] = __byte_perm_S (w0[0], 0, selector);
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 9:
- w0[2] = w0[2] | append0[0] << 8;
- w0[3] = append0[0] >> 24 | append0[1] << 8;
- w1[0] = append0[1] >> 24 | append0[2] << 8;
- w1[1] = append0[2] >> 24 | append0[3] << 8;
- w1[2] = append0[3] >> 24 | append1[0] << 8;
- w1[3] = append1[0] >> 24 | append1[1] << 8;
- w2[0] = append1[1] >> 24 | append1[2] << 8;
- w2[1] = append1[2] >> 24 | append1[3] << 8;
- w2[2] = append1[3] >> 24 | append2[0] << 8;
- w2[3] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w1[0], w0[3], selector);
+ w3[0] = __byte_perm_S (w0[3], w0[2], selector);
+ w2[3] = __byte_perm_S (w0[2], w0[1], selector);
+ w2[2] = __byte_perm_S (w0[1], w0[0], selector);
+ w2[1] = __byte_perm_S (w0[0], 0, selector);
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 10:
- w0[2] = w0[2] | append0[0] << 16;
- w0[3] = append0[0] >> 16 | append0[1] << 16;
- w1[0] = append0[1] >> 16 | append0[2] << 16;
- w1[1] = append0[2] >> 16 | append0[3] << 16;
- w1[2] = append0[3] >> 16 | append1[0] << 16;
- w1[3] = append1[0] >> 16 | append1[1] << 16;
- w2[0] = append1[1] >> 16 | append1[2] << 16;
- w2[1] = append1[2] >> 16 | append1[3] << 16;
- w2[2] = append1[3] >> 16 | append2[0] << 16;
- w2[3] = append2[0] >> 16;
+ w3[1] = __byte_perm_S (w0[3], w0[2], selector);
+ w3[0] = __byte_perm_S (w0[2], w0[1], selector);
+ w2[3] = __byte_perm_S (w0[1], w0[0], selector);
+ w2[2] = __byte_perm_S (w0[0], 0, selector);
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 11:
- w0[2] = w0[2] | append0[0] << 24;
- w0[3] = append0[0] >> 8 | append0[1] << 24;
- w1[0] = append0[1] >> 8 | append0[2] << 24;
- w1[1] = append0[2] >> 8 | append0[3] << 24;
- w1[2] = append0[3] >> 8 | append1[0] << 24;
- w1[3] = append1[0] >> 8 | append1[1] << 24;
- w2[0] = append1[1] >> 8 | append1[2] << 24;
- w2[1] = append1[2] >> 8 | append1[3] << 24;
- w2[2] = append1[3] >> 8 | append2[0] << 24;
- w2[3] = append2[0] >> 8;
+ w3[1] = __byte_perm_S (w0[2], w0[1], selector);
+ w3[0] = __byte_perm_S (w0[1], w0[0], selector);
+ w2[3] = __byte_perm_S (w0[0], 0, selector);
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 12:
- w0[3] = append0[0];
- w1[0] = append0[1];
- w1[1] = append0[2];
- w1[2] = append0[3];
- w1[3] = append1[0];
- w2[0] = append1[1];
- w2[1] = append1[2];
- w2[2] = append1[3];
- w2[3] = append2[0];
+ w3[1] = __byte_perm_S (w0[1], w0[0], selector);
+ w3[0] = __byte_perm_S (w0[0], 0, selector);
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
case 13:
- w0[3] = w0[3] | append0[0] << 8;
- w1[0] = append0[0] >> 24 | append0[1] << 8;
- w1[1] = append0[1] >> 24 | append0[2] << 8;
- w1[2] = append0[2] >> 24 | append0[3] << 8;
- w1[3] = append0[3] >> 24 | append1[0] << 8;
- w2[0] = append1[0] >> 24 | append1[1] << 8;
- w2[1] = append1[1] >> 24 | append1[2] << 8;
- w2[2] = append1[2] >> 24 | append1[3] << 8;
- w2[3] = append1[3] >> 24 | append2[0] << 8;
- w3[0] = append2[0] >> 24;
+ w3[1] = __byte_perm_S (w0[0], 0, selector);
+ w3[0] = 0;
+ w2[3] = 0;
+ w2[2] = 0;
+ w2[1] = 0;
+ w2[0] = 0;
+ w1[3] = 0;
+ w1[2] = 0;
+ w1[1] = 0;
+ w1[0] = 0;
+ w0[3] = 0;
+ w0[2] = 0;
+ w0[1] = 0;
+ w0[0] = 0;
break;
+ }
+ #endif
+}
- case 14:
- w0[3] = w0[3] | append0[0] << 16;
- w1[0] = append0[0] >> 16 | append0[1] << 16;
- w1[1] = append0[1] >> 16 | append0[2] << 16;
- w1[2] = append0[2] >> 16 | append0[3] << 16;
- w1[3] = append0[3] >> 16 | append1[0] << 16;
- w2[0] = append1[0] >> 16 | append1[1] << 16;
- w2[1] = append1[1] >> 16 | append1[2] << 16;
- w2[2] = append1[2] >> 16 | append1[3] << 16;
- w2[3] = append1[3] >> 16 | append2[0] << 16;
- w3[0] = append2[0] >> 16;
- break;
+/**
+ * vector functions on scalar types (for inner loop usage)
+ */
- case 15:
- w0[3] = w0[3] | append0[0] << 24;
- w1[0] = append0[0] >> 8 | append0[1] << 24;
- w1[1] = append0[1] >> 8 | append0[2] << 24;
- w1[2] = append0[2] >> 8 | append0[3] << 24;
- w1[3] = append0[3] >> 8 | append1[0] << 24;
- w2[0] = append1[0] >> 8 | append1[1] << 24;
- w2[1] = append1[1] >> 8 | append1[2] << 24;
- w2[2] = append1[2] >> 8 | append1[3] << 24;
- w2[3] = append1[3] >> 8 | append2[0] << 24;
- w3[0] = append2[0] >> 8;
- break;
+#define PACKVS2(sn,vn,e) \
+ sn[0] = vn[0].s##e; \
+ sn[1] = vn[1].s##e;
+
+#define PACKSV2(sn,vn,e) \
+ vn[0].s##e = sn[0]; \
+ vn[1].s##e = sn[1];
+
+#define PACKVS24(s0,s1,v0,v1,e) \
+ PACKVS4 (s0, v0, e); \
+ PACKVS4 (s1, v1, e);
+
+#define PACKSV24(s0,s1,v0,v1,e) \
+ PACKSV4 (s0, v0, e); \
+ PACKSV4 (s1, v1, e);
+
+#define PACKVS4(sn,vn,e) \
+ sn[0] = vn[0].s##e; \
+ sn[1] = vn[1].s##e; \
+ sn[2] = vn[2].s##e; \
+ sn[3] = vn[3].s##e;
+
+#define PACKSV4(sn,vn,e) \
+ vn[0].s##e = sn[0]; \
+ vn[1].s##e = sn[1]; \
+ vn[2].s##e = sn[2]; \
+ vn[3].s##e = sn[3];
+
+#define PACKVS44(s0,s1,s2,s3,v0,v1,v2,v3,e) \
+ PACKVS4 (s0, v0, e); \
+ PACKVS4 (s1, v1, e); \
+ PACKVS4 (s2, v2, e); \
+ PACKVS4 (s3, v3, e);
+
+#define PACKSV44(s0,s1,s2,s3,v0,v1,v2,v3,e) \
+ PACKSV4 (s0, v0, e); \
+ PACKSV4 (s1, v1, e); \
+ PACKSV4 (s2, v2, e); \
+ PACKSV4 (s3, v3, e);
+
+inline void switch_buffer_by_offset_le_VV (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x offset)
+{
+ #if VECT_SIZE == 1
- case 16:
- w1[0] = append0[0];
- w1[1] = append0[1];
- w1[2] = append0[2];
- w1[3] = append0[3];
- w2[0] = append1[0];
- w2[1] = append1[1];
- w2[2] = append1[2];
- w2[3] = append1[3];
- w3[0] = append2[0];
- break;
+ switch_buffer_by_offset_le_S (w0, w1, w2, w3, offset);
- case 17:
- w1[0] = w1[0] | append0[0] << 8;
- w1[1] = append0[0] >> 24 | append0[1] << 8;
- w1[2] = append0[1] >> 24 | append0[2] << 8;
- w1[3] = append0[2] >> 24 | append0[3] << 8;
- w2[0] = append0[3] >> 24 | append1[0] << 8;
- w2[1] = append1[0] >> 24 | append1[1] << 8;
- w2[2] = append1[1] >> 24 | append1[2] << 8;
- w2[3] = append1[2] >> 24 | append1[3] << 8;
- w3[0] = append1[3] >> 24 | append2[0] << 8;
- w3[1] = append2[0] >> 24;
- break;
+ #else
- case 18:
- w1[0] = w1[0] | append0[0] << 16;
- w1[1] = append0[0] >> 16 | append0[1] << 16;
- w1[2] = append0[1] >> 16 | append0[2] << 16;
- w1[3] = append0[2] >> 16 | append0[3] << 16;
- w2[0] = append0[3] >> 16 | append1[0] << 16;
- w2[1] = append1[0] >> 16 | append1[1] << 16;
- w2[2] = append1[1] >> 16 | append1[2] << 16;
- w2[3] = append1[2] >> 16 | append1[3] << 16;
- w3[0] = append1[3] >> 16 | append2[0] << 16;
- w3[1] = append2[0] >> 16;
- break;
+ u32 t0[4];
+ u32 t1[4];
+ u32 t2[4];
+ u32 t3[4];
- case 19:
- w1[0] = w1[0] | append0[0] << 24;
- w1[1] = append0[0] >> 8 | append0[1] << 24;
- w1[2] = append0[1] >> 8 | append0[2] << 24;
- w1[3] = append0[2] >> 8 | append0[3] << 24;
- w2[0] = append0[3] >> 8 | append1[0] << 24;
- w2[1] = append1[0] >> 8 | append1[1] << 24;
- w2[2] = append1[1] >> 8 | append1[2] << 24;
- w2[3] = append1[2] >> 8 | append1[3] << 24;
- w3[0] = append1[3] >> 8 | append2[0] << 24;
- w3[1] = append2[0] >> 8;
- break;
+ #endif
- case 20:
- w1[1] = append0[0];
- w1[2] = append0[1];
- w1[3] = append0[2];
- w2[0] = append0[3];
- w2[1] = append1[0];
- w2[2] = append1[1];
- w2[3] = append1[2];
- w3[0] = append1[3];
- w3[1] = append2[0];
- break;
+ #if VECT_SIZE == 2
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+
+ #elif VECT_SIZE == 4
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+
+ #elif VECT_SIZE == 8
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 4); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s4); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 4);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 5); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s5); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 5);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 6); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s6); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 6);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 7); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s7); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 7);
+
+ #elif VECT_SIZE == 16
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 4); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s4); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 4);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 5); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s5); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 5);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 6); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s6); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 6);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 7); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s7); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 7);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 8); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s8); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 8);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 9); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.s9); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 9);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, a); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.sa); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, a);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, b); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.sb); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, b);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, c); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.sc); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, c);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, d); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.sd); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, d);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, e); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.se); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, e);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, f); switch_buffer_by_offset_le_S (t0, t1, t2, t3, offset.sf); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, f);
- case 21:
- w1[1] = w1[1] | append0[0] << 8;
- w1[2] = append0[0] >> 24 | append0[1] << 8;
- w1[3] = append0[1] >> 24 | append0[2] << 8;
- w2[0] = append0[2] >> 24 | append0[3] << 8;
- w2[1] = append0[3] >> 24 | append1[0] << 8;
- w2[2] = append1[0] >> 24 | append1[1] << 8;
- w2[3] = append1[1] >> 24 | append1[2] << 8;
- w3[0] = append1[2] >> 24 | append1[3] << 8;
- w3[1] = append1[3] >> 24 | append2[0] << 8;
- break;
+ #endif
+}
- case 22:
- w1[1] = w1[1] | append0[0] << 16;
- w1[2] = append0[0] >> 16 | append0[1] << 16;
- w1[3] = append0[1] >> 16 | append0[2] << 16;
- w2[0] = append0[2] >> 16 | append0[3] << 16;
- w2[1] = append0[3] >> 16 | append1[0] << 16;
- w2[2] = append1[0] >> 16 | append1[1] << 16;
- w2[3] = append1[1] >> 16 | append1[2] << 16;
- w3[0] = append1[2] >> 16 | append1[3] << 16;
- w3[1] = append1[3] >> 16 | append2[0] << 16;
- break;
+inline void append_0x01_2x4_VV (u32x w0[4], u32x w1[4], const u32x offset)
+{
+ #if VECT_SIZE == 1
- case 23:
- w1[1] = w1[1] | append0[0] << 24;
- w1[2] = append0[0] >> 8 | append0[1] << 24;
- w1[3] = append0[1] >> 8 | append0[2] << 24;
- w2[0] = append0[2] >> 8 | append0[3] << 24;
- w2[1] = append0[3] >> 8 | append1[0] << 24;
- w2[2] = append1[0] >> 8 | append1[1] << 24;
- w2[3] = append1[1] >> 8 | append1[2] << 24;
- w3[0] = append1[2] >> 8 | append1[3] << 24;
- w3[1] = append1[3] >> 8 | append2[0] << 24;
- break;
+ append_0x01_2x4_S (w0, w1, offset);
- case 24:
- w1[2] = append0[0];
- w1[3] = append0[1];
- w2[0] = append0[2];
- w2[1] = append0[3];
- w2[2] = append1[0];
- w2[3] = append1[1];
- w3[0] = append1[2];
- w3[1] = append1[3];
- break;
+ #else
- case 25:
- w1[2] = w1[2] | append0[0] << 8;
- w1[3] = append0[0] >> 24 | append0[1] << 8;
- w2[0] = append0[1] >> 24 | append0[2] << 8;
- w2[1] = append0[2] >> 24 | append0[3] << 8;
- w2[2] = append0[3] >> 24 | append1[0] << 8;
- w2[3] = append1[0] >> 24 | append1[1] << 8;
- w3[0] = append1[1] >> 24 | append1[2] << 8;
- w3[1] = append1[2] >> 24 | append1[3] << 8;
- break;
+ u32 t0[4];
+ u32 t1[4];
- case 26:
- w1[2] = w1[2] | append0[0] << 16;
- w1[3] = append0[0] >> 16 | append0[1] << 16;
- w2[0] = append0[1] >> 16 | append0[2] << 16;
- w2[1] = append0[2] >> 16 | append0[3] << 16;
- w2[2] = append0[3] >> 16 | append1[0] << 16;
- w2[3] = append1[0] >> 16 | append1[1] << 16;
- w3[0] = append1[1] >> 16 | append1[2] << 16;
- w3[1] = append1[2] >> 16 | append1[3] << 16;
- break;
+ #endif
- case 27:
- w1[2] = w1[2] | append0[0] << 24;
- w1[3] = append0[0] >> 8 | append0[1] << 24;
- w2[0] = append0[1] >> 8 | append0[2] << 24;
- w2[1] = append0[2] >> 8 | append0[3] << 24;
- w2[2] = append0[3] >> 8 | append1[0] << 24;
- w2[3] = append1[0] >> 8 | append1[1] << 24;
- w3[0] = append1[1] >> 8 | append1[2] << 24;
- w3[1] = append1[2] >> 8 | append1[3] << 24;
- break;
+ #if VECT_SIZE == 2
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x01_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x01_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+
+ #elif VECT_SIZE == 4
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x01_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x01_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x01_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x01_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+
+ #elif VECT_SIZE == 8
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x01_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x01_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x01_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x01_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+ PACKVS24 (t0, t1, w0, w1, 4); append_0x01_2x4_S (t0, t1, offset.s4); PACKSV24 (t0, t1, w0, w1, 4);
+ PACKVS24 (t0, t1, w0, w1, 5); append_0x01_2x4_S (t0, t1, offset.s5); PACKSV24 (t0, t1, w0, w1, 5);
+ PACKVS24 (t0, t1, w0, w1, 6); append_0x01_2x4_S (t0, t1, offset.s6); PACKSV24 (t0, t1, w0, w1, 6);
+ PACKVS24 (t0, t1, w0, w1, 7); append_0x01_2x4_S (t0, t1, offset.s7); PACKSV24 (t0, t1, w0, w1, 7);
+
+ #elif VECT_SIZE == 16
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x01_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x01_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x01_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x01_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+ PACKVS24 (t0, t1, w0, w1, 4); append_0x01_2x4_S (t0, t1, offset.s4); PACKSV24 (t0, t1, w0, w1, 4);
+ PACKVS24 (t0, t1, w0, w1, 5); append_0x01_2x4_S (t0, t1, offset.s5); PACKSV24 (t0, t1, w0, w1, 5);
+ PACKVS24 (t0, t1, w0, w1, 6); append_0x01_2x4_S (t0, t1, offset.s6); PACKSV24 (t0, t1, w0, w1, 6);
+ PACKVS24 (t0, t1, w0, w1, 7); append_0x01_2x4_S (t0, t1, offset.s7); PACKSV24 (t0, t1, w0, w1, 7);
+ PACKVS24 (t0, t1, w0, w1, 8); append_0x01_2x4_S (t0, t1, offset.s8); PACKSV24 (t0, t1, w0, w1, 8);
+ PACKVS24 (t0, t1, w0, w1, 9); append_0x01_2x4_S (t0, t1, offset.s9); PACKSV24 (t0, t1, w0, w1, 9);
+ PACKVS24 (t0, t1, w0, w1, a); append_0x01_2x4_S (t0, t1, offset.sa); PACKSV24 (t0, t1, w0, w1, a);
+ PACKVS24 (t0, t1, w0, w1, b); append_0x01_2x4_S (t0, t1, offset.sb); PACKSV24 (t0, t1, w0, w1, b);
+ PACKVS24 (t0, t1, w0, w1, c); append_0x01_2x4_S (t0, t1, offset.sc); PACKSV24 (t0, t1, w0, w1, c);
+ PACKVS24 (t0, t1, w0, w1, d); append_0x01_2x4_S (t0, t1, offset.sd); PACKSV24 (t0, t1, w0, w1, d);
+ PACKVS24 (t0, t1, w0, w1, e); append_0x01_2x4_S (t0, t1, offset.se); PACKSV24 (t0, t1, w0, w1, e);
+ PACKVS24 (t0, t1, w0, w1, f); append_0x01_2x4_S (t0, t1, offset.sf); PACKSV24 (t0, t1, w0, w1, f);
- case 28:
- w1[3] = append0[0];
- w2[0] = append0[1];
- w2[1] = append0[2];
- w2[2] = append0[3];
- w2[3] = append1[0];
- w3[0] = append1[1];
- w3[1] = append1[2];
- break;
+ #endif
+}
- case 29:
- w1[3] = w1[3] | append0[0] << 8;
- w2[0] = append0[0] >> 24 | append0[1] << 8;
- w2[1] = append0[1] >> 24 | append0[2] << 8;
- w2[2] = append0[2] >> 24 | append0[3] << 8;
- w2[3] = append0[3] >> 24 | append1[0] << 8;
- w3[0] = append1[0] >> 24 | append1[1] << 8;
- w3[1] = append1[1] >> 24 | append1[2] << 8;
- break;
+inline void append_0x80_2x4_VV (u32x w0[4], u32x w1[4], const u32x offset)
+{
+ #if VECT_SIZE == 1
- case 30:
- w1[3] = w1[3] | append0[0] << 16;
- w2[0] = append0[0] >> 16 | append0[1] << 16;
- w2[1] = append0[1] >> 16 | append0[2] << 16;
- w2[2] = append0[2] >> 16 | append0[3] << 16;
- w2[3] = append0[3] >> 16 | append1[0] << 16;
- w3[0] = append1[0] >> 16 | append1[1] << 16;
- w3[1] = append1[1] >> 16 | append1[2] << 16;
- break;
+ append_0x80_2x4_S (w0, w1, offset);
- case 31:
- w1[3] = w1[3] | append0[0] << 24;
- w2[0] = append0[0] >> 8 | append0[1] << 24;
- w2[1] = append0[1] >> 8 | append0[2] << 24;
- w2[2] = append0[2] >> 8 | append0[3] << 24;
- w2[3] = append0[3] >> 8 | append1[0] << 24;
- w3[0] = append1[0] >> 8 | append1[1] << 24;
- w3[1] = append1[1] >> 8 | append1[2] << 24;
- break;
+ #else
- case 32:
- w2[0] = append0[0];
- w2[1] = append0[1];
- w2[2] = append0[2];
- w2[3] = append0[3];
- w3[0] = append1[0];
- w3[1] = append1[1];
- break;
- }
+ u32 t0[4];
+ u32 t1[4];
+
+ #endif
+
+ #if VECT_SIZE == 2
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x80_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x80_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+
+ #elif VECT_SIZE == 4
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x80_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x80_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x80_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x80_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+
+ #elif VECT_SIZE == 8
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x80_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x80_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x80_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x80_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+ PACKVS24 (t0, t1, w0, w1, 4); append_0x80_2x4_S (t0, t1, offset.s4); PACKSV24 (t0, t1, w0, w1, 4);
+ PACKVS24 (t0, t1, w0, w1, 5); append_0x80_2x4_S (t0, t1, offset.s5); PACKSV24 (t0, t1, w0, w1, 5);
+ PACKVS24 (t0, t1, w0, w1, 6); append_0x80_2x4_S (t0, t1, offset.s6); PACKSV24 (t0, t1, w0, w1, 6);
+ PACKVS24 (t0, t1, w0, w1, 7); append_0x80_2x4_S (t0, t1, offset.s7); PACKSV24 (t0, t1, w0, w1, 7);
+
+ #elif VECT_SIZE == 16
+
+ PACKVS24 (t0, t1, w0, w1, 0); append_0x80_2x4_S (t0, t1, offset.s0); PACKSV24 (t0, t1, w0, w1, 0);
+ PACKVS24 (t0, t1, w0, w1, 1); append_0x80_2x4_S (t0, t1, offset.s1); PACKSV24 (t0, t1, w0, w1, 1);
+ PACKVS24 (t0, t1, w0, w1, 2); append_0x80_2x4_S (t0, t1, offset.s2); PACKSV24 (t0, t1, w0, w1, 2);
+ PACKVS24 (t0, t1, w0, w1, 3); append_0x80_2x4_S (t0, t1, offset.s3); PACKSV24 (t0, t1, w0, w1, 3);
+ PACKVS24 (t0, t1, w0, w1, 4); append_0x80_2x4_S (t0, t1, offset.s4); PACKSV24 (t0, t1, w0, w1, 4);
+ PACKVS24 (t0, t1, w0, w1, 5); append_0x80_2x4_S (t0, t1, offset.s5); PACKSV24 (t0, t1, w0, w1, 5);
+ PACKVS24 (t0, t1, w0, w1, 6); append_0x80_2x4_S (t0, t1, offset.s6); PACKSV24 (t0, t1, w0, w1, 6);
+ PACKVS24 (t0, t1, w0, w1, 7); append_0x80_2x4_S (t0, t1, offset.s7); PACKSV24 (t0, t1, w0, w1, 7);
+ PACKVS24 (t0, t1, w0, w1, 8); append_0x80_2x4_S (t0, t1, offset.s8); PACKSV24 (t0, t1, w0, w1, 8);
+ PACKVS24 (t0, t1, w0, w1, 9); append_0x80_2x4_S (t0, t1, offset.s9); PACKSV24 (t0, t1, w0, w1, 9);
+ PACKVS24 (t0, t1, w0, w1, a); append_0x80_2x4_S (t0, t1, offset.sa); PACKSV24 (t0, t1, w0, w1, a);
+ PACKVS24 (t0, t1, w0, w1, b); append_0x80_2x4_S (t0, t1, offset.sb); PACKSV24 (t0, t1, w0, w1, b);
+ PACKVS24 (t0, t1, w0, w1, c); append_0x80_2x4_S (t0, t1, offset.sc); PACKSV24 (t0, t1, w0, w1, c);
+ PACKVS24 (t0, t1, w0, w1, d); append_0x80_2x4_S (t0, t1, offset.sd); PACKSV24 (t0, t1, w0, w1, d);
+ PACKVS24 (t0, t1, w0, w1, e); append_0x80_2x4_S (t0, t1, offset.se); PACKSV24 (t0, t1, w0, w1, e);
+ PACKVS24 (t0, t1, w0, w1, f); append_0x80_2x4_S (t0, t1, offset.sf); PACKSV24 (t0, t1, w0, w1, f);
+
+ #endif
}
-*/
+inline void append_0x80_4x4_VV (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x offset)
+{
+ #if VECT_SIZE == 1
+
+ append_0x80_4x4_S (w0, w1, w2, w3, offset);
+
+ #else
+
+ u32 t0[4];
+ u32 t1[4];
+ u32 t2[4];
+ u32 t3[4];
+
+ #endif
+
+ #if VECT_SIZE == 2
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); append_0x80_4x4_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); append_0x80_4x4_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+
+ #elif VECT_SIZE == 4
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); append_0x80_4x4_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); append_0x80_4x4_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); append_0x80_4x4_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); append_0x80_4x4_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+
+ #elif VECT_SIZE == 8
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); append_0x80_4x4_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); append_0x80_4x4_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); append_0x80_4x4_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); append_0x80_4x4_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 4); append_0x80_4x4_S (t0, t1, t2, t3, offset.s4); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 4);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 5); append_0x80_4x4_S (t0, t1, t2, t3, offset.s5); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 5);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 6); append_0x80_4x4_S (t0, t1, t2, t3, offset.s6); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 6);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 7); append_0x80_4x4_S (t0, t1, t2, t3, offset.s7); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 7);
+
+ #elif VECT_SIZE == 16
+
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 0); append_0x80_4x4_S (t0, t1, t2, t3, offset.s0); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 0);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 1); append_0x80_4x4_S (t0, t1, t2, t3, offset.s1); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 1);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 2); append_0x80_4x4_S (t0, t1, t2, t3, offset.s2); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 2);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 3); append_0x80_4x4_S (t0, t1, t2, t3, offset.s3); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 3);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 4); append_0x80_4x4_S (t0, t1, t2, t3, offset.s4); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 4);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 5); append_0x80_4x4_S (t0, t1, t2, t3, offset.s5); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 5);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 6); append_0x80_4x4_S (t0, t1, t2, t3, offset.s6); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 6);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 7); append_0x80_4x4_S (t0, t1, t2, t3, offset.s7); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 7);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 8); append_0x80_4x4_S (t0, t1, t2, t3, offset.s8); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 8);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, 9); append_0x80_4x4_S (t0, t1, t2, t3, offset.s9); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, 9);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, a); append_0x80_4x4_S (t0, t1, t2, t3, offset.sa); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, a);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, b); append_0x80_4x4_S (t0, t1, t2, t3, offset.sb); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, b);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, c); append_0x80_4x4_S (t0, t1, t2, t3, offset.sc); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, c);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, d); append_0x80_4x4_S (t0, t1, t2, t3, offset.sd); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, d);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, e); append_0x80_4x4_S (t0, t1, t2, t3, offset.se); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, e);
+ PACKVS44 (t0, t1, t2, t3, w0, w1, w2, w3, f); append_0x80_4x4_S (t0, t1, t2, t3, offset.sf); PACKSV44 (t0, t1, t2, t3, w0, w1, w2, w3, f);
+
+ #endif
+}
projects / hashcat.git / blobdiff