#define _SHA1_
-#define NEW_SIMD_CODE
-
#include "include/constants.h"
#include "include/kernel_vendor.h"
#include "include/kernel_functions.c"
#include "OpenCL/types_ocl.c"
#include "OpenCL/common.c"
-#include "OpenCL/simd.c"
+
+#define COMPARE_S "OpenCL/check_single_comp4.c"
+#define COMPARE_M "OpenCL/check_multi_comp4.c"
// no unicode yet
if (gid >= gid_max) return;
- u32 pws0[4] = { 0 };
- u32 pws1[4] = { 0 };
+ u32 wordl0[4];
+
+ wordl0[0] = pws[gid].i[ 0];
+ wordl0[1] = pws[gid].i[ 1];
+ wordl0[2] = pws[gid].i[ 2];
+ wordl0[3] = pws[gid].i[ 3];
+
+ u32 wordl1[4];
+
+ wordl1[0] = pws[gid].i[ 4];
+ wordl1[1] = pws[gid].i[ 5];
+ wordl1[2] = pws[gid].i[ 6];
+ wordl1[3] = pws[gid].i[ 7];
+
+ u32 wordl2[4];
- pws0[0] = pws[gid].i[0];
- pws0[1] = pws[gid].i[1];
- pws0[2] = pws[gid].i[2];
- pws0[3] = pws[gid].i[3];
- pws1[0] = pws[gid].i[4];
- pws1[1] = pws[gid].i[5];
- pws1[2] = pws[gid].i[6];
- pws1[3] = pws[gid].i[7];
+ wordl2[0] = 0;
+ wordl2[1] = 0;
+ wordl2[2] = 0;
+ wordl2[3] = 0;
+
+ u32 wordl3[4];
+
+ wordl3[0] = 0;
+ wordl3[1] = 0;
+ wordl3[2] = 0;
+ wordl3[3] = 0;
const u32 pw_l_len = pws[gid].pw_len;
+ if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
+ {
+ switch_buffer_by_offset_le (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
+ }
+
/**
* salt
*/
u32 salt_buf0[4];
- u32 salt_buf1[4];
salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0];
salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1];
salt_buf0[2] = salt_bufs[salt_pos].salt_buf[2];
salt_buf0[3] = salt_bufs[salt_pos].salt_buf[3];
+
+ u32 salt_buf1[4];
+
salt_buf1[0] = salt_bufs[salt_pos].salt_buf[4];
salt_buf1[1] = salt_bufs[salt_pos].salt_buf[5];
salt_buf1[2] = salt_bufs[salt_pos].salt_buf[6];
* loop
*/
- for (u32 il_pos = 0; il_pos < combs_cnt; il_pos += VECT_SIZE)
+ for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
{
- const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
+ const u32 pw_r_len = combs_buf[il_pos].pw_len;
+
+ const u32 pw_len = pw_l_len + pw_r_len;
+
+ u32 wordr0[4];
- const u32x pw_len = pw_l_len + pw_r_len;
+ wordr0[0] = combs_buf[il_pos].i[0];
+ wordr0[1] = combs_buf[il_pos].i[1];
+ wordr0[2] = combs_buf[il_pos].i[2];
+ wordr0[3] = combs_buf[il_pos].i[3];
- u32x wordr0[4] = { 0 };
- u32x wordr1[4] = { 0 };
- u32x wordr2[4] = { 0 };
- u32x wordr3[4] = { 0 };
+ u32 wordr1[4];
- wordr0[0] = ix_create_combt (combs_buf, il_pos, 0);
- wordr0[1] = ix_create_combt (combs_buf, il_pos, 1);
- wordr0[2] = ix_create_combt (combs_buf, il_pos, 2);
- wordr0[3] = ix_create_combt (combs_buf, il_pos, 3);
- wordr1[0] = ix_create_combt (combs_buf, il_pos, 4);
- wordr1[1] = ix_create_combt (combs_buf, il_pos, 5);
- wordr1[2] = ix_create_combt (combs_buf, il_pos, 6);
- wordr1[3] = ix_create_combt (combs_buf, il_pos, 7);
+ wordr1[0] = combs_buf[il_pos].i[4];
+ wordr1[1] = combs_buf[il_pos].i[5];
+ wordr1[2] = combs_buf[il_pos].i[6];
+ wordr1[3] = combs_buf[il_pos].i[7];
+
+ u32 wordr2[4];
+
+ wordr2[0] = 0;
+ wordr2[1] = 0;
+ wordr2[2] = 0;
+ wordr2[3] = 0;
+
+ u32 wordr3[4];
+
+ wordr3[0] = 0;
+ wordr3[1] = 0;
+ wordr3[2] = 0;
+ wordr3[3] = 0;
if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
{
const u32 pw_salt_len = (pw_len * 2) + salt_len;
- u32x w0[4];
- u32x w1[4];
- u32x w2[4];
- u32x w3[4];
+ u32 w0[4];
+ u32 w1[4];
+ u32 w2[4];
+ u32 w3[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w3[2] = 0;
w3[3] = 0;
- u32x w0_t[4];
- u32x w1_t[4];
- u32x w2_t[4];
- u32x w3_t[4];
+ u32 w0_t[4];
+ u32 w1_t[4];
+ u32 w2_t[4];
+ u32 w3_t[4];
make_unicode (w0, w0_t, w1_t);
make_unicode (w1, w2_t, w3_t);
//w3_t[2] = swap32 (w3_t[2]);
//w3_t[3] = swap32 (w3_t[3]);
- u32x a = SHA1M_A;
- u32x b = SHA1M_B;
- u32x c = SHA1M_C;
- u32x d = SHA1M_D;
- u32x e = SHA1M_E;
+ u32 a = SHA1M_A;
+ u32 b = SHA1M_B;
+ u32 c = SHA1M_C;
+ u32 d = SHA1M_D;
+ u32 e = SHA1M_E;
#undef K
#define K SHA1C00
w3_t[2] = rotl32 ((w2_t[3] ^ w1_t[2] ^ w0_t[0] ^ w3_t[2]), 1u); SHA1_STEP (SHA1_F1, c, d, e, a, b, w3_t[2]);
w3_t[3] = rotl32 ((w3_t[0] ^ w1_t[3] ^ w0_t[1] ^ w3_t[3]), 1u); SHA1_STEP (SHA1_F1, b, c, d, e, a, w3_t[3]);
- COMPARE_M_SIMD (d, e, c, b);
+
+ const u32 r0 = d;
+ const u32 r1 = e;
+ const u32 r2 = c;
+ const u32 r3 = b;
+
+ #include COMPARE_M
}
}
if (gid >= gid_max) return;
- u32 pws0[4] = { 0 };
- u32 pws1[4] = { 0 };
+ u32 wordl0[4];
+
+ wordl0[0] = pws[gid].i[ 0];
+ wordl0[1] = pws[gid].i[ 1];
+ wordl0[2] = pws[gid].i[ 2];
+ wordl0[3] = pws[gid].i[ 3];
+
+ u32 wordl1[4];
+
+ wordl1[0] = pws[gid].i[ 4];
+ wordl1[1] = pws[gid].i[ 5];
+ wordl1[2] = pws[gid].i[ 6];
+ wordl1[3] = pws[gid].i[ 7];
+
+ u32 wordl2[4];
+
+ wordl2[0] = 0;
+ wordl2[1] = 0;
+ wordl2[2] = 0;
+ wordl2[3] = 0;
- pws0[0] = pws[gid].i[0];
- pws0[1] = pws[gid].i[1];
- pws0[2] = pws[gid].i[2];
- pws0[3] = pws[gid].i[3];
- pws1[0] = pws[gid].i[4];
- pws1[1] = pws[gid].i[5];
- pws1[2] = pws[gid].i[6];
- pws1[3] = pws[gid].i[7];
+ u32 wordl3[4];
+
+ wordl3[0] = 0;
+ wordl3[1] = 0;
+ wordl3[2] = 0;
+ wordl3[3] = 0;
const u32 pw_l_len = pws[gid].pw_len;
+ if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
+ {
+ switch_buffer_by_offset_le (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
+ }
+
/**
* salt
*/
u32 salt_buf0[4];
- u32 salt_buf1[4];
salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0];
salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1];
salt_buf0[2] = salt_bufs[salt_pos].salt_buf[2];
salt_buf0[3] = salt_bufs[salt_pos].salt_buf[3];
+
+ u32 salt_buf1[4];
+
salt_buf1[0] = salt_bufs[salt_pos].salt_buf[4];
salt_buf1[1] = salt_bufs[salt_pos].salt_buf[5];
salt_buf1[2] = salt_bufs[salt_pos].salt_buf[6];
* reverse
*/
- const u32 e_rev = rotl32_S (search[1], 2u);
+ const u32 e_rev = rotl32 (search[1], 2u);
/**
* loop
*/
- for (u32 il_pos = 0; il_pos < combs_cnt; il_pos += VECT_SIZE)
+ for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
{
- const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
+ const u32 pw_r_len = combs_buf[il_pos].pw_len;
+
+ const u32 pw_len = pw_l_len + pw_r_len;
+
+ u32 wordr0[4];
+
+ wordr0[0] = combs_buf[il_pos].i[0];
+ wordr0[1] = combs_buf[il_pos].i[1];
+ wordr0[2] = combs_buf[il_pos].i[2];
+ wordr0[3] = combs_buf[il_pos].i[3];
- const u32x pw_len = pw_l_len + pw_r_len;
+ u32 wordr1[4];
- u32x wordr0[4] = { 0 };
- u32x wordr1[4] = { 0 };
- u32x wordr2[4] = { 0 };
- u32x wordr3[4] = { 0 };
+ wordr1[0] = combs_buf[il_pos].i[4];
+ wordr1[1] = combs_buf[il_pos].i[5];
+ wordr1[2] = combs_buf[il_pos].i[6];
+ wordr1[3] = combs_buf[il_pos].i[7];
- wordr0[0] = ix_create_combt (combs_buf, il_pos, 0);
- wordr0[1] = ix_create_combt (combs_buf, il_pos, 1);
- wordr0[2] = ix_create_combt (combs_buf, il_pos, 2);
- wordr0[3] = ix_create_combt (combs_buf, il_pos, 3);
- wordr1[0] = ix_create_combt (combs_buf, il_pos, 4);
- wordr1[1] = ix_create_combt (combs_buf, il_pos, 5);
- wordr1[2] = ix_create_combt (combs_buf, il_pos, 6);
- wordr1[3] = ix_create_combt (combs_buf, il_pos, 7);
+ u32 wordr2[4];
+
+ wordr2[0] = 0;
+ wordr2[1] = 0;
+ wordr2[2] = 0;
+ wordr2[3] = 0;
+
+ u32 wordr3[4];
+
+ wordr3[0] = 0;
+ wordr3[1] = 0;
+ wordr3[2] = 0;
+ wordr3[3] = 0;
if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
{
const u32 pw_salt_len = (pw_len * 2) + salt_len;
- u32x w0[4];
- u32x w1[4];
- u32x w2[4];
- u32x w3[4];
+ u32 w0[4];
+ u32 w1[4];
+ u32 w2[4];
+ u32 w3[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w3[2] = 0;
w3[3] = 0;
- u32x w0_t[4];
- u32x w1_t[4];
- u32x w2_t[4];
- u32x w3_t[4];
+ u32 w0_t[4];
+ u32 w1_t[4];
+ u32 w2_t[4];
+ u32 w3_t[4];
make_unicode (w0, w0_t, w1_t);
make_unicode (w1, w2_t, w3_t);
//w3_t[2] = swap32 (w3_t[2]);
//w3_t[3] = swap32 (w3_t[3]);
- u32x a = SHA1M_A;
- u32x b = SHA1M_B;
- u32x c = SHA1M_C;
- u32x d = SHA1M_D;
- u32x e = SHA1M_E;
+ u32 a = SHA1M_A;
+ u32 b = SHA1M_B;
+ u32 c = SHA1M_C;
+ u32 d = SHA1M_D;
+ u32 e = SHA1M_E;
#undef K
#define K SHA1C00
w2_t[3] = rotl32 ((w2_t[0] ^ w0_t[3] ^ w3_t[1] ^ w2_t[3]), 1u); SHA1_STEP (SHA1_F1, a, b, c, d, e, w2_t[3]);
w3_t[0] = rotl32 ((w2_t[1] ^ w1_t[0] ^ w3_t[2] ^ w3_t[0]), 1u); SHA1_STEP (SHA1_F1, e, a, b, c, d, w3_t[0]);
- if (MATCHES_NONE_VS (e, e_rev)) continue;
+ if (allx (e != e_rev)) continue;
w3_t[1] = rotl32 ((w2_t[2] ^ w1_t[1] ^ w3_t[3] ^ w3_t[1]), 1u); SHA1_STEP (SHA1_F1, d, e, a, b, c, w3_t[1]);
w3_t[2] = rotl32 ((w2_t[3] ^ w1_t[2] ^ w0_t[0] ^ w3_t[2]), 1u); SHA1_STEP (SHA1_F1, c, d, e, a, b, w3_t[2]);
w3_t[3] = rotl32 ((w3_t[0] ^ w1_t[3] ^ w0_t[1] ^ w3_t[3]), 1u); SHA1_STEP (SHA1_F1, b, c, d, e, a, w3_t[3]);
- COMPARE_S_SIMD (d, e, c, b);
+
+ const u32 r0 = d;
+ const u32 r1 = e;
+ const u32 r2 = c;
+ const u32 r3 = b;
+
+ #include COMPARE_S
}
}