X-Git-Url: https://www.flypig.org.uk/git/?a=blobdiff_plain;f=OpenCL%2Fm09820_a1.cl;h=dd2a05a3084be42cfeaddec69c30d42cf4fda028;hb=b409e5e9e162927b25af88e672326b246f17ec5f;hp=73f62b9fbe54fbe26df23e690f22f6bc91111716;hpb=6c10ca58533d9da85e561b339b027492ce1794d3;p=hashcat.git

diff --git a/OpenCL/m09820_a1.cl b/OpenCL/m09820_a1.cl
index 73f62b9..dd2a05a 100644
--- a/OpenCL/m09820_a1.cl
+++ b/OpenCL/m09820_a1.cl
@@ -5,6 +5,8 @@
 
 #define _OLDOFFICE34_
 
+#define NEW_SIMD_CODE
+
 #include "include/constants.h"
 #include "include/kernel_vendor.h"
 
@@ -16,9 +18,7 @@
 #include "include/kernel_functions.c"
 #include "OpenCL/types_ocl.c"
 #include "OpenCL/common.c"
-
-#define COMPARE_S "OpenCL/check_single_comp4.c"
-#define COMPARE_M "OpenCL/check_multi_comp4.c"
+#include "OpenCL/simd.c"
 
 static void sha1_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[5])
 {
@@ -164,41 +164,20 @@ __kernel void m09820_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
   if (gid >= gid_max) return;
 
-  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;
-
-  u32 wordl3[4];
+  u32 pws0[4] = { 0 };
+  u32 pws1[4] = { 0 };
 
-  wordl3[0] = 0;
-  wordl3[1] = 0;
-  wordl3[2] = 0;
-  wordl3[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];
 
   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
    */
@@ -216,69 +195,55 @@ __kernel void m09820_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
    * loop
    */
 
-  for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < combs_cnt; il_pos += VECT_SIZE)
   {
-    const u32 pw_r_len = combs_buf[il_pos].pw_len;
+    const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
 
-    const u32 pw_len = pw_l_len + pw_r_len;
+    const u32x pw_len = pw_l_len + pw_r_len;
 
     const u32 pw_salt_len = (pw_len * 2) + salt_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];
-
-    u32 wordr1[4];
-
-    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;
+    u32x wordr0[4] = { 0 };
+    u32x wordr1[4] = { 0 };
+    u32x wordr2[4] = { 0 };
+    u32x wordr3[4] = { 0 };
 
-    u32 wordr3[4];
-
-    wordr3[0] = 0;
-    wordr3[1] = 0;
-    wordr3[2] = 0;
-    wordr3[3] = 0;
+    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);
 
     if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
     {
       switch_buffer_by_offset_le (wordr0, wordr1, wordr2, wordr3, pw_l_len);
     }
 
-    u32 w0[4];
+    u32x w0[4];
 
     w0[0] = wordl0[0] | wordr0[0];
     w0[1] = wordl0[1] | wordr0[1];
     w0[2] = wordl0[2] | wordr0[2];
     w0[3] = wordl0[3] | wordr0[3];
 
-    u32 w1[4];
+    u32x w1[4];
 
     w1[0] = wordl1[0] | wordr1[0];
     w1[1] = wordl1[1] | wordr1[1];
     w1[2] = wordl1[2] | wordr1[2];
     w1[3] = wordl1[3] | wordr1[3];
 
-    u32 w2[4];
+    u32x w2[4];
 
     w2[0] = wordl2[0] | wordr2[0];
     w2[1] = wordl2[1] | wordr2[1];
     w2[2] = wordl2[2] | wordr2[2];
     w2[3] = wordl2[3] | wordr2[3];
 
-    u32 w3[4];
+    u32x w3[4];
 
     w3[0] = wordl3[0] | wordr3[0];
     w3[1] = wordl3[1] | wordr3[1];
@@ -287,10 +252,10 @@ __kernel void m09820_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     append_0x80_2x4 (w0, w1, pw_len);
 
-    u32 w0_t[4];
-    u32 w1_t[4];
-    u32 w2_t[4];
-    u32 w3_t[4];
+    u32x w0_t[4];
+    u32x w1_t[4];
+    u32x w2_t[4];
+    u32x w3_t[4];
 
     make_unicode (w0, w0_t, w1_t);
     make_unicode (w1, w2_t, w3_t);
@@ -349,8 +314,8 @@ __kernel void m09820_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (w0_t, w1_t, w2_t, w3_t, digest);
 
-    u32 a = swap32 (digest[0]);
-    u32 b = swap32 (digest[1]) & 0xff;
+    u32x a = swap32 (digest[0]);
+    u32x b = swap32 (digest[1]) & 0xff;
 
     const u32 r0 = a;
     const u32 r1 = b;
@@ -385,41 +350,20 @@ __kernel void m09820_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
   if (gid >= gid_max) return;
 
-  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 pws0[4] = { 0 };
+  u32 pws1[4] = { 0 };
 
-  u32 wordl2[4];
-
-  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;
+  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];
 
   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);
-  }
-
   /**
    * digest
    */
@@ -449,69 +393,55 @@ __kernel void m09820_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
    * loop
    */
 
-  for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < combs_cnt; il_pos += VECT_SIZE)
   {
-    const u32 pw_r_len = combs_buf[il_pos].pw_len;
+    const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
 
-    const u32 pw_len = pw_l_len + pw_r_len;
+    const u32x pw_len = pw_l_len + pw_r_len;
 
     const u32 pw_salt_len = (pw_len * 2) + salt_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];
-
-    u32 wordr1[4];
-
-    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];
+    u32x wordr0[4] = { 0 };
+    u32x wordr1[4] = { 0 };
+    u32x wordr2[4] = { 0 };
+    u32x wordr3[4] = { 0 };
 
-    wordr3[0] = 0;
-    wordr3[1] = 0;
-    wordr3[2] = 0;
-    wordr3[3] = 0;
+    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);
 
     if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
     {
       switch_buffer_by_offset_le (wordr0, wordr1, wordr2, wordr3, pw_l_len);
     }
 
-    u32 w0[4];
+    u32x w0[4];
 
     w0[0] = wordl0[0] | wordr0[0];
     w0[1] = wordl0[1] | wordr0[1];
     w0[2] = wordl0[2] | wordr0[2];
     w0[3] = wordl0[3] | wordr0[3];
 
-    u32 w1[4];
+    u32x w1[4];
 
     w1[0] = wordl1[0] | wordr1[0];
     w1[1] = wordl1[1] | wordr1[1];
     w1[2] = wordl1[2] | wordr1[2];
     w1[3] = wordl1[3] | wordr1[3];
 
-    u32 w2[4];
+    u32x w2[4];
 
     w2[0] = wordl2[0] | wordr2[0];
     w2[1] = wordl2[1] | wordr2[1];
     w2[2] = wordl2[2] | wordr2[2];
     w2[3] = wordl2[3] | wordr2[3];
 
-    u32 w3[4];
+    u32x w3[4];
 
     w3[0] = wordl3[0] | wordr3[0];
     w3[1] = wordl3[1] | wordr3[1];
@@ -520,10 +450,10 @@ __kernel void m09820_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     append_0x80_2x4 (w0, w1, pw_len);
 
-    u32 w0_t[4];
-    u32 w1_t[4];
-    u32 w2_t[4];
-    u32 w3_t[4];
+    u32x w0_t[4];
+    u32x w1_t[4];
+    u32x w2_t[4];
+    u32x w3_t[4];
 
     make_unicode (w0, w0_t, w1_t);
     make_unicode (w1, w2_t, w3_t);
@@ -582,8 +512,8 @@ __kernel void m09820_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (w0_t, w1_t, w2_t, w3_t, digest);
 
-    u32 a = swap32 (digest[0]);
-    u32 b = swap32 (digest[1]) & 0xff;
+    u32x a = swap32 (digest[0]);
+    u32x b = swap32 (digest[1]) & 0xff;
 
     const u32 r0 = a;
     const u32 r1 = b;