Converted _a3 kernels, use SIMD for CPU and GPU

[hashcat.git] / OpenCL / m00160_a3.cl

diff --git a/OpenCL/m00160_a3.cl b/OpenCL/m00160_a3.cl
index 4f3a34a..126927b 100644 (file)
--- a/OpenCL/m00160_a3.cl
+++ b/OpenCL/m00160_a3.cl
@@ -5,6 +5,8 @@
 
 #define _SHA1_
 
+#define NEW_SIMD_CODE
+
 #include "include/constants.h"
 #include "include/kernel_vendor.h"
 
@@ -16,34 +18,32 @@
 #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"
-
-static void sha1_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[5])
+static void sha1_transform (const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4], u32x digest[5])
 {
-  u32 A = digest[0];
-  u32 B = digest[1];
-  u32 C = digest[2];
-  u32 D = digest[3];
-  u32 E = digest[4];
-
-  u32 w0_t = w0[0];
-  u32 w1_t = w0[1];
-  u32 w2_t = w0[2];
-  u32 w3_t = w0[3];
-  u32 w4_t = w1[0];
-  u32 w5_t = w1[1];
-  u32 w6_t = w1[2];
-  u32 w7_t = w1[3];
-  u32 w8_t = w2[0];
-  u32 w9_t = w2[1];
-  u32 wa_t = w2[2];
-  u32 wb_t = w2[3];
-  u32 wc_t = w3[0];
-  u32 wd_t = w3[1];
-  u32 we_t = w3[2];
-  u32 wf_t = w3[3];
+  u32x A = digest[0];
+  u32x B = digest[1];
+  u32x C = digest[2];
+  u32x D = digest[3];
+  u32x E = digest[4];
+
+  u32x w0_t = w0[0];
+  u32x w1_t = w0[1];
+  u32x w2_t = w0[2];
+  u32x w3_t = w0[3];
+  u32x w4_t = w1[0];
+  u32x w5_t = w1[1];
+  u32x w6_t = w1[2];
+  u32x w7_t = w1[3];
+  u32x w8_t = w2[0];
+  u32x w9_t = w2[1];
+  u32x wa_t = w2[2];
+  u32x wb_t = w2[3];
+  u32x wc_t = w3[0];
+  u32x wd_t = w3[1];
+  u32x we_t = w3[2];
+  u32x wf_t = w3[3];
 
   #undef K
   #define K SHA1C00
@@ -148,7 +148,7 @@ static void sha1_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], c
   digest[4] += E;
 }
 
-static void hmac_sha1_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[5], u32 opad[5])
+static void hmac_sha1_pad (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x ipad[5], u32x opad[5])
 {
   w0[0] = w0[0] ^ 0x36363636;
   w0[1] = w0[1] ^ 0x36363636;
@@ -201,7 +201,7 @@ static void hmac_sha1_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[
   sha1_transform (w0, w1, w2, w3, opad);
 }
 
-static void hmac_sha1_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[5], u32 opad[5], u32 digest[5])
+static void hmac_sha1_run (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x ipad[5], u32x opad[5], u32x digest[5])
 {
   digest[0] = ipad[0];
   digest[1] = ipad[1];
@@ -268,36 +268,36 @@ static void m00160m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
    * pads
    */
 
-  u32 w0_t[4];
+  u32x w0_t[4];
 
   w0_t[0] = swap32 (salt_buf0[0]);
   w0_t[1] = swap32 (salt_buf0[1]);
   w0_t[2] = swap32 (salt_buf0[2]);
   w0_t[3] = swap32 (salt_buf0[3]);
 
-  u32 w1_t[4];
+  u32x w1_t[4];
 
   w1_t[0] = swap32 (salt_buf1[0]);
   w1_t[1] = swap32 (salt_buf1[1]);
   w1_t[2] = swap32 (salt_buf1[2]);
   w1_t[3] = swap32 (salt_buf1[3]);
 
-  u32 w2_t[4];
+  u32x w2_t[4];
 
   w2_t[0] = 0;
   w2_t[1] = 0;
   w2_t[2] = 0;
   w2_t[3] = 0;
 
-  u32 w3_t[4];
+  u32x w3_t[4];
 
   w3_t[0] = 0;
   w3_t[1] = 0;
   w3_t[2] = 0;
   w3_t[3] = 0;
 
-  u32 ipad[5];
-  u32 opad[5];
+  u32x ipad[5];
+  u32x opad[5];
 
   hmac_sha1_pad (w0_t, w1_t, w2_t, w3_t, ipad, opad);
 
@@ -307,13 +307,13 @@ static void m00160m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
 
   u32 w0l = w0[0];
 
-  for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos += VECT_SIZE)
   {
-    const u32 w0r = bfs_buf[il_pos].i;
+    const u32x w0r = w0r_create_bft (bfs_buf, il_pos);
 
-    w0[0] = w0l | w0r;
+    const u32x w0lr = w0l | w0r;
 
-    w0_t[0] = w0[0];
+    w0_t[0] = w0lr;
     w0_t[1] = w0[1];
     w0_t[2] = w0[2];
     w0_t[3] = w0[3];
@@ -330,16 +330,11 @@ static void m00160m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
     w3_t[2] = 0;
     w3_t[3] = (64 + pw_len) * 8;
 
-    u32 digest[5];
+    u32x digest[5];
 
     hmac_sha1_run (w0_t, w1_t, w2_t, w3_t, ipad, opad, digest);
 
-    const u32 r0 = digest[3];
-    const u32 r1 = digest[4];
-    const u32 r2 = digest[2];
-    const u32 r3 = digest[1];
-
-    #include COMPARE_M
+    COMPARE_M_SIMD (digest[3], digest[4], digest[2], digest[1]);
   }
 }
 
@@ -374,36 +369,36 @@ static void m00160s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
    * pads
    */
 
-  u32 w0_t[4];
+  u32x w0_t[4];
 
   w0_t[0] = swap32 (salt_buf0[0]);
   w0_t[1] = swap32 (salt_buf0[1]);
   w0_t[2] = swap32 (salt_buf0[2]);
   w0_t[3] = swap32 (salt_buf0[3]);
 
-  u32 w1_t[4];
+  u32x w1_t[4];
 
   w1_t[0] = swap32 (salt_buf1[0]);
   w1_t[1] = swap32 (salt_buf1[1]);
   w1_t[2] = swap32 (salt_buf1[2]);
   w1_t[3] = swap32 (salt_buf1[3]);
 
-  u32 w2_t[4];
+  u32x w2_t[4];
 
   w2_t[0] = 0;
   w2_t[1] = 0;
   w2_t[2] = 0;
   w2_t[3] = 0;
 
-  u32 w3_t[4];
+  u32x w3_t[4];
 
   w3_t[0] = 0;
   w3_t[1] = 0;
   w3_t[2] = 0;
   w3_t[3] = 0;
 
-  u32 ipad[5];
-  u32 opad[5];
+  u32x ipad[5];
+  u32x opad[5];
 
   hmac_sha1_pad (w0_t, w1_t, w2_t, w3_t, ipad, opad);
 
@@ -425,13 +420,13 @@ static void m00160s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
 
   u32 w0l = w0[0];
 
-  for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos += VECT_SIZE)
   {
-    const u32 w0r = bfs_buf[il_pos].i;
+    const u32x w0r = w0r_create_bft (bfs_buf, il_pos);
 
-    w0[0] = w0l | w0r;
+    const u32x w0lr = w0l | w0r;
 
-    w0_t[0] = w0[0];
+    w0_t[0] = w0lr;
     w0_t[1] = w0[1];
     w0_t[2] = w0[2];
     w0_t[3] = w0[3];
@@ -448,16 +443,11 @@ static void m00160s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
     w3_t[2] = 0;
     w3_t[3] = (64 + pw_len) * 8;
 
-    u32 digest[5];
+    u32x digest[5];
 
     hmac_sha1_run (w0_t, w1_t, w2_t, w3_t, ipad, opad, digest);
 
-    const u32 r0 = digest[3];
-    const u32 r1 = digest[4];
-    const u32 r2 = digest[2];
-    const u32 r3 = digest[1];
-
-    #include COMPARE_S
+    COMPARE_S_SIMD (digest[3], digest[4], digest[2], digest[1]);
   }
 }