From 7b10348f7b9f86bb9e57ee93a3baca50c93213c4 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Sun, 28 Feb 2016 19:58:09 +0100
Subject: [PATCH] New SIMD code for -a 1 -m 1440

---
 OpenCL/m01440_a1.cl | 489 ++++++++++++++++++++++----------------------
 1 file changed, 244 insertions(+), 245 deletions(-)

diff --git a/OpenCL/m01440_a1.cl b/OpenCL/m01440_a1.cl
index c1e7995e7..47fc1f97c 100644
--- a/OpenCL/m01440_a1.cl
+++ b/OpenCL/m01440_a1.cl
@@ -5,6 +5,8 @@
 
 #define _SHA256_
 
+#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"
 
 __kernel void m01440_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __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, const u32 combs_mode, const u32 gid_max)
 {
@@ -36,58 +36,45 @@ __kernel void m01440_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
   if (gid >= gid_max) return;
 
-  u32 wordl0[4];
+  u32 pw_buf0[4];
+  u32 pw_buf1[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];
-
-  wordl3[0] = 0;
-  wordl3[1] = 0;
-  wordl3[2] = 0;
-  wordl3[3] = 0;
+  pw_buf0[0] = pws[gid].i[0];
+  pw_buf0[1] = pws[gid].i[1];
+  pw_buf0[2] = pws[gid].i[2];
+  pw_buf0[3] = pws[gid].i[3];
+  pw_buf1[0] = pws[gid].i[4];
+  pw_buf1[1] = pws[gid].i[5];
+  pw_buf1[2] = pws[gid].i[6];
+  pw_buf1[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
    */
 
   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];
+  u32 salt_buf2[4];
+  u32 salt_buf3[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];
+  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];
+  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];
+  salt_buf2[0] = salt_bufs[salt_pos].salt_buf[ 8];
+  salt_buf2[1] = salt_bufs[salt_pos].salt_buf[ 9];
+  salt_buf2[2] = salt_bufs[salt_pos].salt_buf[10];
+  salt_buf2[3] = salt_bufs[salt_pos].salt_buf[11];
+  salt_buf3[0] = salt_bufs[salt_pos].salt_buf[12];
+  salt_buf3[1] = salt_bufs[salt_pos].salt_buf[13];
+  salt_buf3[2] = salt_bufs[salt_pos].salt_buf[14];
+  salt_buf3[3] = salt_bufs[salt_pos].salt_buf[15];
 
   const u32 salt_len = salt_bufs[salt_pos].salt_len;
 
@@ -95,43 +82,57 @@ __kernel void m01440_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
    * loop
    */
 
-  for (u32 il_pos = 0; il_pos < il_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < il_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;
 
-    u32 wordr0[4];
-    u32 wordr1[4];
-    u32 wordr2[4];
-    u32 wordr3[4];
+    /**
+     * concat password candidate
+     */
 
-    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];
-    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];
-    wordr2[0] = 0;
-    wordr2[1] = 0;
-    wordr2[2] = 0;
-    wordr2[3] = 0;
-    wordr3[0] = 0;
-    wordr3[1] = 0;
-    wordr3[2] = 0;
-    wordr3[3] = 0;
+    u32x wordl0[4] = { 0 };
+    u32x wordl1[4] = { 0 };
+    u32x wordl2[4] = { 0 };
+    u32x wordl3[4] = { 0 };
+
+    wordl0[0] = pw_buf0[0];
+    wordl0[1] = pw_buf0[1];
+    wordl0[2] = pw_buf0[2];
+    wordl0[3] = pw_buf0[3];
+    wordl1[0] = pw_buf1[0];
+    wordl1[1] = pw_buf1[1];
+    wordl1[2] = pw_buf1[2];
+    wordl1[3] = pw_buf1[3];
+
+    u32x wordr0[4] = { 0 };
+    u32x wordr1[4] = { 0 };
+    u32x wordr2[4] = { 0 };
+    u32x wordr3[4] = { 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);
+      switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+    }
+    else
+    {
+      switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
     }
 
-    u32 w0[4];
-    u32 w1[4];
-    u32 w2[4];
-    u32 w3[4];
+    u32x w0[4];
+    u32x w1[4];
+    u32x w2[4];
+    u32x w3[4];
 
     w0[0] = wordl0[0] | wordr0[0];
     w0[1] = wordl0[1] | wordr0[1];
@@ -150,62 +151,65 @@ __kernel void m01440_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
     w3[2] = wordl3[2] | wordr3[2];
     w3[3] = wordl3[3] | wordr3[3];
 
+    make_unicode (w1, w2, w3);
+    make_unicode (w0, w0, w1);
+
+    const u32x pw_len2 = pw_len * 2;
+
     /**
      * prepend salt
      */
 
-    const u32 pw_salt_len = (pw_len * 2) + salt_len;
+    switch_buffer_by_offset_le (w0, w1, w2, w3, salt_len);
 
-    u32 w0_t2[4];
-    u32 w1_t2[4];
-    u32 w2_t2[4];
-    u32 w3_t2[4];
+    const u32x pw_salt_len = pw_len2 + salt_len;
 
-    make_unicode (w0, w0_t2, w1_t2);
-    make_unicode (w1, w2_t2, w3_t2);
-
-    switch_buffer_by_offset_le (w0_t2, w1_t2, w2_t2, w3_t2, salt_len);
-
-    w0_t2[0] |= salt_buf0[0];
-    w0_t2[1] |= salt_buf0[1];
-    w0_t2[2] |= salt_buf0[2];
-    w0_t2[3] |= salt_buf0[3];
-    w1_t2[0] |= salt_buf1[0];
-    w1_t2[1] |= salt_buf1[1];
-    w1_t2[2] |= salt_buf1[2];
-    w1_t2[3] |= salt_buf1[3];
-
-    append_0x80_4x4 (w0_t2, w1_t2, w2_t2, w3_t2, pw_salt_len);
+    w0[0] |= salt_buf0[0];
+    w0[1] |= salt_buf0[1];
+    w0[2] |= salt_buf0[2];
+    w0[3] |= salt_buf0[3];
+    w1[0] |= salt_buf1[0];
+    w1[1] |= salt_buf1[1];
+    w1[2] |= salt_buf1[2];
+    w1[3] |= salt_buf1[3];
+    w2[0] |= salt_buf2[0];
+    w2[1] |= salt_buf2[1];
+    w2[2] |= salt_buf2[2];
+    w2[3] |= salt_buf2[3];
+    w3[0] |= salt_buf3[0];
+    w3[1] |= salt_buf3[1];
+    w3[2] |= salt_buf3[2];
+    w3[3] |= salt_buf3[3];
 
     /**
-     * sha256
+     * sha1
      */
 
-    u32 w0_t = swap32 (w0_t2[0]);
-    u32 w1_t = swap32 (w0_t2[1]);
-    u32 w2_t = swap32 (w0_t2[2]);
-    u32 w3_t = swap32 (w0_t2[3]);
-    u32 w4_t = swap32 (w1_t2[0]);
-    u32 w5_t = swap32 (w1_t2[1]);
-    u32 w6_t = swap32 (w1_t2[2]);
-    u32 w7_t = swap32 (w1_t2[3]);
-    u32 w8_t = swap32 (w2_t2[0]);
-    u32 w9_t = swap32 (w2_t2[1]);
-    u32 wa_t = swap32 (w2_t2[2]);
-    u32 wb_t = swap32 (w2_t2[3]);
-    u32 wc_t = swap32 (w3_t2[0]);
-    u32 wd_t = swap32 (w3_t2[1]);
-    u32 we_t = 0;
-    u32 wf_t = pw_salt_len * 8;
+    u32x w0_t = swap32 (w0[0]);
+    u32x w1_t = swap32 (w0[1]);
+    u32x w2_t = swap32 (w0[2]);
+    u32x w3_t = swap32 (w0[3]);
+    u32x w4_t = swap32 (w1[0]);
+    u32x w5_t = swap32 (w1[1]);
+    u32x w6_t = swap32 (w1[2]);
+    u32x w7_t = swap32 (w1[3]);
+    u32x w8_t = swap32 (w2[0]);
+    u32x w9_t = swap32 (w2[1]);
+    u32x wa_t = swap32 (w2[2]);
+    u32x wb_t = swap32 (w2[3]);
+    u32x wc_t = swap32 (w3[0]);
+    u32x wd_t = swap32 (w3[1]);
+    u32x we_t = 0;
+    u32x wf_t = pw_salt_len * 8;
 
-    u32 a = SHA256M_A;
-    u32 b = SHA256M_B;
-    u32 c = SHA256M_C;
-    u32 d = SHA256M_D;
-    u32 e = SHA256M_E;
-    u32 f = SHA256M_F;
-    u32 g = SHA256M_G;
-    u32 h = SHA256M_H;
+    u32x a = SHA256M_A;
+    u32x b = SHA256M_B;
+    u32x c = SHA256M_C;
+    u32x d = SHA256M_D;
+    u32x e = SHA256M_E;
+    u32x f = SHA256M_F;
+    u32x g = SHA256M_G;
+    u32x h = SHA256M_H;
 
     SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, SHA256C00);
     SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, SHA256C01);
@@ -275,13 +279,7 @@ __kernel void m01440_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
     we_t = SHA256_EXPAND (wc_t, w7_t, wf_t, we_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, SHA256C3e);
     wf_t = SHA256_EXPAND (wd_t, w8_t, w0_t, wf_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, SHA256C3f);
 
-
-    const u32 r0 = d;
-    const u32 r1 = h;
-    const u32 r2 = c;
-    const u32 r3 = g;
-
-    #include COMPARE_M
+    COMPARE_M_SIMD (d, h, c, g);
   }
 }
 
@@ -309,58 +307,45 @@ __kernel void m01440_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
   if (gid >= gid_max) return;
 
-  u32 wordl0[4];
+  u32 pw_buf0[4];
+  u32 pw_buf1[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];
-
-  wordl3[0] = 0;
-  wordl3[1] = 0;
-  wordl3[2] = 0;
-  wordl3[3] = 0;
+  pw_buf0[0] = pws[gid].i[0];
+  pw_buf0[1] = pws[gid].i[1];
+  pw_buf0[2] = pws[gid].i[2];
+  pw_buf0[3] = pws[gid].i[3];
+  pw_buf1[0] = pws[gid].i[4];
+  pw_buf1[1] = pws[gid].i[5];
+  pw_buf1[2] = pws[gid].i[6];
+  pw_buf1[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
    */
 
   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];
+  u32 salt_buf2[4];
+  u32 salt_buf3[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];
+  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];
+  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];
+  salt_buf2[0] = salt_bufs[salt_pos].salt_buf[ 8];
+  salt_buf2[1] = salt_bufs[salt_pos].salt_buf[ 9];
+  salt_buf2[2] = salt_bufs[salt_pos].salt_buf[10];
+  salt_buf2[3] = salt_bufs[salt_pos].salt_buf[11];
+  salt_buf3[0] = salt_bufs[salt_pos].salt_buf[12];
+  salt_buf3[1] = salt_bufs[salt_pos].salt_buf[13];
+  salt_buf3[2] = salt_bufs[salt_pos].salt_buf[14];
+  salt_buf3[3] = salt_bufs[salt_pos].salt_buf[15];
 
   const u32 salt_len = salt_bufs[salt_pos].salt_len;
 
@@ -380,43 +365,57 @@ __kernel void m01440_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
    * loop
    */
 
-  for (u32 il_pos = 0; il_pos < il_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < il_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;
 
-    u32 wordr0[4];
-    u32 wordr1[4];
-    u32 wordr2[4];
-    u32 wordr3[4];
+    /**
+     * concat password candidate
+     */
 
-    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];
-    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];
-    wordr2[0] = 0;
-    wordr2[1] = 0;
-    wordr2[2] = 0;
-    wordr2[3] = 0;
-    wordr3[0] = 0;
-    wordr3[1] = 0;
-    wordr3[2] = 0;
-    wordr3[3] = 0;
+    u32x wordl0[4] = { 0 };
+    u32x wordl1[4] = { 0 };
+    u32x wordl2[4] = { 0 };
+    u32x wordl3[4] = { 0 };
+
+    wordl0[0] = pw_buf0[0];
+    wordl0[1] = pw_buf0[1];
+    wordl0[2] = pw_buf0[2];
+    wordl0[3] = pw_buf0[3];
+    wordl1[0] = pw_buf1[0];
+    wordl1[1] = pw_buf1[1];
+    wordl1[2] = pw_buf1[2];
+    wordl1[3] = pw_buf1[3];
+
+    u32x wordr0[4] = { 0 };
+    u32x wordr1[4] = { 0 };
+    u32x wordr2[4] = { 0 };
+    u32x wordr3[4] = { 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);
+      switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+    }
+    else
+    {
+      switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
     }
 
-    u32 w0[4];
-    u32 w1[4];
-    u32 w2[4];
-    u32 w3[4];
+    u32x w0[4];
+    u32x w1[4];
+    u32x w2[4];
+    u32x w3[4];
 
     w0[0] = wordl0[0] | wordr0[0];
     w0[1] = wordl0[1] | wordr0[1];
@@ -435,62 +434,65 @@ __kernel void m01440_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
     w3[2] = wordl3[2] | wordr3[2];
     w3[3] = wordl3[3] | wordr3[3];
 
+    make_unicode (w1, w2, w3);
+    make_unicode (w0, w0, w1);
+
+    const u32x pw_len2 = pw_len * 2;
+
     /**
      * prepend salt
      */
 
-    const u32 pw_salt_len = (pw_len * 2) + salt_len;
+    switch_buffer_by_offset_le (w0, w1, w2, w3, salt_len);
 
-    u32 w0_t2[4];
-    u32 w1_t2[4];
-    u32 w2_t2[4];
-    u32 w3_t2[4];
+    const u32x pw_salt_len = pw_len2 + salt_len;
 
-    make_unicode (w0, w0_t2, w1_t2);
-    make_unicode (w1, w2_t2, w3_t2);
-
-    switch_buffer_by_offset_le (w0_t2, w1_t2, w2_t2, w3_t2, salt_len);
-
-    w0_t2[0] |= salt_buf0[0];
-    w0_t2[1] |= salt_buf0[1];
-    w0_t2[2] |= salt_buf0[2];
-    w0_t2[3] |= salt_buf0[3];
-    w1_t2[0] |= salt_buf1[0];
-    w1_t2[1] |= salt_buf1[1];
-    w1_t2[2] |= salt_buf1[2];
-    w1_t2[3] |= salt_buf1[3];
-
-    append_0x80_4x4 (w0_t2, w1_t2, w2_t2, w3_t2, pw_salt_len);
+    w0[0] |= salt_buf0[0];
+    w0[1] |= salt_buf0[1];
+    w0[2] |= salt_buf0[2];
+    w0[3] |= salt_buf0[3];
+    w1[0] |= salt_buf1[0];
+    w1[1] |= salt_buf1[1];
+    w1[2] |= salt_buf1[2];
+    w1[3] |= salt_buf1[3];
+    w2[0] |= salt_buf2[0];
+    w2[1] |= salt_buf2[1];
+    w2[2] |= salt_buf2[2];
+    w2[3] |= salt_buf2[3];
+    w3[0] |= salt_buf3[0];
+    w3[1] |= salt_buf3[1];
+    w3[2] |= salt_buf3[2];
+    w3[3] |= salt_buf3[3];
 
     /**
-     * sha256
+     * sha1
      */
 
-    u32 w0_t = swap32 (w0_t2[0]);
-    u32 w1_t = swap32 (w0_t2[1]);
-    u32 w2_t = swap32 (w0_t2[2]);
-    u32 w3_t = swap32 (w0_t2[3]);
-    u32 w4_t = swap32 (w1_t2[0]);
-    u32 w5_t = swap32 (w1_t2[1]);
-    u32 w6_t = swap32 (w1_t2[2]);
-    u32 w7_t = swap32 (w1_t2[3]);
-    u32 w8_t = swap32 (w2_t2[0]);
-    u32 w9_t = swap32 (w2_t2[1]);
-    u32 wa_t = swap32 (w2_t2[2]);
-    u32 wb_t = swap32 (w2_t2[3]);
-    u32 wc_t = swap32 (w3_t2[0]);
-    u32 wd_t = swap32 (w3_t2[1]);
-    u32 we_t = 0;
-    u32 wf_t = pw_salt_len * 8;
+    u32x w0_t = swap32 (w0[0]);
+    u32x w1_t = swap32 (w0[1]);
+    u32x w2_t = swap32 (w0[2]);
+    u32x w3_t = swap32 (w0[3]);
+    u32x w4_t = swap32 (w1[0]);
+    u32x w5_t = swap32 (w1[1]);
+    u32x w6_t = swap32 (w1[2]);
+    u32x w7_t = swap32 (w1[3]);
+    u32x w8_t = swap32 (w2[0]);
+    u32x w9_t = swap32 (w2[1]);
+    u32x wa_t = swap32 (w2[2]);
+    u32x wb_t = swap32 (w2[3]);
+    u32x wc_t = swap32 (w3[0]);
+    u32x wd_t = swap32 (w3[1]);
+    u32x we_t = 0;
+    u32x wf_t = pw_salt_len * 8;
 
-    u32 a = SHA256M_A;
-    u32 b = SHA256M_B;
-    u32 c = SHA256M_C;
-    u32 d = SHA256M_D;
-    u32 e = SHA256M_E;
-    u32 f = SHA256M_F;
-    u32 g = SHA256M_G;
-    u32 h = SHA256M_H;
+    u32x a = SHA256M_A;
+    u32x b = SHA256M_B;
+    u32x c = SHA256M_C;
+    u32x d = SHA256M_D;
+    u32x e = SHA256M_E;
+    u32x f = SHA256M_F;
+    u32x g = SHA256M_G;
+    u32x h = SHA256M_H;
 
     SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, SHA256C00);
     SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, SHA256C01);
@@ -556,17 +558,14 @@ __kernel void m01440_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
     wa_t = SHA256_EXPAND (w8_t, w3_t, wb_t, wa_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, wa_t, SHA256C3a);
     wb_t = SHA256_EXPAND (w9_t, w4_t, wc_t, wb_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, wb_t, SHA256C3b);
     wc_t = SHA256_EXPAND (wa_t, w5_t, wd_t, wc_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, wc_t, SHA256C3c);
+
+    if (MATCHES_NONE_VS (d, search[0])) continue;
+
     wd_t = SHA256_EXPAND (wb_t, w6_t, we_t, wd_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, wd_t, SHA256C3d);
     we_t = SHA256_EXPAND (wc_t, w7_t, wf_t, we_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, SHA256C3e);
     wf_t = SHA256_EXPAND (wd_t, w8_t, w0_t, wf_t); SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, SHA256C3f);
 
-
-    const u32 r0 = d;
-    const u32 r1 = h;
-    const u32 r2 = c;
-    const u32 r3 = g;
-
-    #include COMPARE_S
+    COMPARE_S_SIMD (d, h, c, g);
   }
 }