Remove barrier from DCC1, high-end maxwells didn't like it

[hashcat.git] / OpenCL / m01100_a3.cl

diff --git a/OpenCL/m01100_a3.cl b/OpenCL/m01100_a3.cl
index ab67bb7..793930b 100644 (file)
--- a/OpenCL/m01100_a3.cl
+++ b/OpenCL/m01100_a3.cl
@@ -20,7 +20,7 @@
 #include "OpenCL/common.c"
 #include "OpenCL/simd.c"
 
-static void m01100m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
+void m01100m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
 {
   /**
    * modifier
@@ -33,28 +33,24 @@ static void m01100m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
    * salt
    */
 
-  u32 salt_buf0[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];
-  salt_buf1[3] = salt_bufs[salt_pos].salt_buf[ 7];
-
-  u32 salt_buf2[4];
-
-  salt_buf2[0] = salt_bufs[salt_pos].salt_buf[ 8];
-  salt_buf2[1] = salt_bufs[salt_pos].salt_buf[ 9];
-  salt_buf2[2] = 0;
-  salt_buf2[3] = 0;
-
-  const u32 salt_len = salt_bufs[salt_pos].salt_len;
+  #define salt_buf00 salt_bufs[salt_pos].salt_buf[ 0]
+  #define salt_buf01 salt_bufs[salt_pos].salt_buf[ 1]
+  #define salt_buf02 salt_bufs[salt_pos].salt_buf[ 2]
+  #define salt_buf03 salt_bufs[salt_pos].salt_buf[ 3]
+  #define salt_buf04 salt_bufs[salt_pos].salt_buf[ 4]
+  #define salt_buf05 salt_bufs[salt_pos].salt_buf[ 5]
+  #define salt_buf06 salt_bufs[salt_pos].salt_buf[ 6]
+  #define salt_buf07 salt_bufs[salt_pos].salt_buf[ 7]
+  #define salt_buf08 salt_bufs[salt_pos].salt_buf[ 8]
+  #define salt_buf09 salt_bufs[salt_pos].salt_buf[ 9]
+  #define salt_buf10 salt_bufs[salt_pos].salt_buf[10]
+  #define salt_buf11 salt_bufs[salt_pos].salt_buf[11]
+  #define salt_buf12 salt_bufs[salt_pos].salt_buf[12]
+  #define salt_buf13 salt_bufs[salt_pos].salt_buf[13]
+  #define salt_buf14 salt_bufs[salt_pos].salt_buf[14]
+  #define salt_buf15 salt_bufs[salt_pos].salt_buf[15]
+
+  #define salt_len   salt_bufs[salt_pos].salt_len
 
   /**
    * base
@@ -193,16 +189,16 @@ static void m01100m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
     w0_t[1] = b;
     w0_t[2] = c;
     w0_t[3] = d;
-    w1_t[0] = salt_buf0[0];
-    w1_t[1] = salt_buf0[1];
-    w1_t[2] = salt_buf0[2];
-    w1_t[3] = salt_buf0[3];
-    w2_t[0] = salt_buf1[0];
-    w2_t[1] = salt_buf1[1];
-    w2_t[2] = salt_buf1[2];
-    w2_t[3] = salt_buf1[3];
-    w3_t[0] = salt_buf2[0];
-    w3_t[1] = salt_buf2[1];
+    w1_t[0] = salt_buf00;
+    w1_t[1] = salt_buf01;
+    w1_t[2] = salt_buf02;
+    w1_t[3] = salt_buf03;
+    w2_t[0] = salt_buf04;
+    w2_t[1] = salt_buf05;
+    w2_t[2] = salt_buf06;
+    w2_t[3] = salt_buf07;
+    w3_t[0] = salt_buf08;
+    w3_t[1] = salt_buf09;
     w3_t[2] = (16 + salt_len) * 8;
     w3_t[3] = 0;
 
@@ -266,7 +262,7 @@ static void m01100m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
   }
 }
 
-static void m01100s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
+void m01100s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
 {
   /**
    * modifier
@@ -279,28 +275,24 @@ static void m01100s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
    * salt
    */
 
-  u32 salt_buf0[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];
-  salt_buf1[3] = salt_bufs[salt_pos].salt_buf[ 7];
-
-  u32 salt_buf2[4];
-
-  salt_buf2[0] = salt_bufs[salt_pos].salt_buf[ 8];
-  salt_buf2[1] = salt_bufs[salt_pos].salt_buf[ 9];
-  salt_buf2[2] = 0;
-  salt_buf2[3] = 0;
-
-  const u32 salt_len = salt_bufs[salt_pos].salt_len;
+  #define salt_buf00 salt_bufs[salt_pos].salt_buf[ 0]
+  #define salt_buf01 salt_bufs[salt_pos].salt_buf[ 1]
+  #define salt_buf02 salt_bufs[salt_pos].salt_buf[ 2]
+  #define salt_buf03 salt_bufs[salt_pos].salt_buf[ 3]
+  #define salt_buf04 salt_bufs[salt_pos].salt_buf[ 4]
+  #define salt_buf05 salt_bufs[salt_pos].salt_buf[ 5]
+  #define salt_buf06 salt_bufs[salt_pos].salt_buf[ 6]
+  #define salt_buf07 salt_bufs[salt_pos].salt_buf[ 7]
+  #define salt_buf08 salt_bufs[salt_pos].salt_buf[ 8]
+  #define salt_buf09 salt_bufs[salt_pos].salt_buf[ 9]
+  #define salt_buf10 salt_bufs[salt_pos].salt_buf[10]
+  #define salt_buf11 salt_bufs[salt_pos].salt_buf[11]
+  #define salt_buf12 salt_bufs[salt_pos].salt_buf[12]
+  #define salt_buf13 salt_bufs[salt_pos].salt_buf[13]
+  #define salt_buf14 salt_bufs[salt_pos].salt_buf[14]
+  #define salt_buf15 salt_bufs[salt_pos].salt_buf[15]
+
+  #define salt_len   salt_bufs[salt_pos].salt_len
 
   /**
    * base
@@ -451,16 +443,16 @@ static void m01100s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k
     w0_t[1] = b;
     w0_t[2] = c;
     w0_t[3] = d;
-    w1_t[0] = salt_buf0[0];
-    w1_t[1] = salt_buf0[1];
-    w1_t[2] = salt_buf0[2];
-    w1_t[3] = salt_buf0[3];
-    w2_t[0] = salt_buf1[0];
-    w2_t[1] = salt_buf1[1];
-    w2_t[2] = salt_buf1[2];
-    w2_t[3] = salt_buf1[3];
-    w3_t[0] = salt_buf2[0];
-    w3_t[1] = salt_buf2[1];
+    w1_t[0] = salt_buf00;
+    w1_t[1] = salt_buf01;
+    w1_t[2] = salt_buf02;
+    w1_t[3] = salt_buf03;
+    w2_t[0] = salt_buf04;
+    w2_t[1] = salt_buf05;
+    w2_t[2] = salt_buf06;
+    w2_t[3] = salt_buf07;
+    w3_t[0] = salt_buf08;
+    w3_t[1] = salt_buf09;
     w3_t[2] = (16 + salt_len) * 8;
     w3_t[3] = 0;