diff --git a/OpenCL/m07800_a0.cl b/OpenCL/m07800_a0.cl
index 74f919337..f1c3c2d0f 100644
--- a/OpenCL/m07800_a0.cl
+++ b/OpenCL/m07800_a0.cl
@@ -5,6 +5,9 @@
 
 #define _SAPG_
 
+//incompatible data-dependant code
+//#define NEW_SIMD_CODE
+
 #include "include/constants.h"
 #include "include/kernel_vendor.h"
 
@@ -18,9 +21,7 @@
 #include "OpenCL/common.c"
 #include "include/rp_kernel.h"
 #include "OpenCL/rp.c"
-
-#define COMPARE_S "OpenCL/check_single_comp4.c"
-#define COMPARE_M "OpenCL/check_multi_comp4.c"
+#include "OpenCL/simd.c"
 
 #define GETSHIFTEDINT(a,n) amd_bytealign ((a)[((n)/4)+1], (a)[((n)/4)+0], (n))
 
@@ -209,18 +210,16 @@ __kernel void m07800_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
   if (gid >= gid_max) return;
 
   u32 pw_buf0[4];
-
-  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];
-
   u32 pw_buf1[4];
 
-  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];
+  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_len = pws[gid].pw_len;
 
@@ -245,65 +244,36 @@ __kernel void m07800_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)
   {
-    u32 w0[4];
-
-    w0[0] = pw_buf0[0];
-    w0[1] = pw_buf0[1];
-    w0[2] = pw_buf0[2];
-    w0[3] = pw_buf0[3];
-
-    u32 w1[4];
+    u32x w0[4] = { 0 };
+    u32x w1[4] = { 0 };
+    u32x w2[4] = { 0 };
+    u32x w3[4] = { 0 };
 
-    w1[0] = pw_buf1[0];
-    w1[1] = pw_buf1[1];
-    w1[2] = pw_buf1[2];
-    w1[3] = pw_buf1[3];
-
-    u32 w2[4];
-
-    w2[0] = 0;
-    w2[1] = 0;
-    w2[2] = 0;
-    w2[3] = 0;
-
-    u32 w3[4];
-
-    w3[0] = 0;
-    w3[1] = 0;
-    w3[2] = 0;
-    w3[3] = 0;
-
-    const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
+    const u32x out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
 
     /**
-     * append salt
+     * SAP
      */
 
     u32 s0[4];
+    u32 s1[4];
+    u32 s2[4];
+    u32 s3[4];
 
     s0[0] = salt_buf[0];
     s0[1] = salt_buf[1];
     s0[2] = salt_buf[2];
     s0[3] = salt_buf[3];
-
-    u32 s1[4];
-
     s1[0] = salt_buf[4];
     s1[1] = salt_buf[5];
     s1[2] = salt_buf[6];
     s1[3] = salt_buf[7];
-
-    u32 s2[4];
-
     s2[0] = 0;
     s2[1] = 0;
     s2[2] = 0;
     s2[3] = 0;
-
-    u32 s3[4];
-
     s3[0] = 0;
     s3[1] = 0;
     s3[2] = 0;
@@ -443,12 +413,7 @@ __kernel void m07800_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }
 
@@ -477,18 +442,16 @@ __kernel void m07800_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
   if (gid >= gid_max) return;
 
   u32 pw_buf0[4];
-
-  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];
-
   u32 pw_buf1[4];
 
-  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];
+  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_len = pws[gid].pw_len;
 
@@ -525,65 +488,36 @@ __kernel void m07800_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)
   {
-    u32 w0[4];
-
-    w0[0] = pw_buf0[0];
-    w0[1] = pw_buf0[1];
-    w0[2] = pw_buf0[2];
-    w0[3] = pw_buf0[3];
-
-    u32 w1[4];
-
-    w1[0] = pw_buf1[0];
-    w1[1] = pw_buf1[1];
-    w1[2] = pw_buf1[2];
-    w1[3] = pw_buf1[3];
+    u32x w0[4] = { 0 };
+    u32x w1[4] = { 0 };
+    u32x w2[4] = { 0 };
+    u32x w3[4] = { 0 };
 
-    u32 w2[4];
-
-    w2[0] = 0;
-    w2[1] = 0;
-    w2[2] = 0;
-    w2[3] = 0;
-
-    u32 w3[4];
-
-    w3[0] = 0;
-    w3[1] = 0;
-    w3[2] = 0;
-    w3[3] = 0;
-
-    const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
+    const u32x out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
 
     /**
-     * append salt
+     * SAP
      */
 
     u32 s0[4];
+    u32 s1[4];
+    u32 s2[4];
+    u32 s3[4];
 
     s0[0] = salt_buf[0];
     s0[1] = salt_buf[1];
     s0[2] = salt_buf[2];
     s0[3] = salt_buf[3];
-
-    u32 s1[4];
-
     s1[0] = salt_buf[4];
     s1[1] = salt_buf[5];
     s1[2] = salt_buf[6];
     s1[3] = salt_buf[7];
-
-    u32 s2[4];
-
     s2[0] = 0;
     s2[1] = 0;
     s2[2] = 0;
     s2[3] = 0;
-
-    u32 s3[4];
-
     s3[0] = 0;
     s3[1] = 0;
     s3[2] = 0;
@@ -723,12 +657,7 @@ __kernel void m07800_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }
 
diff --git a/OpenCL/m07800_a1.cl b/OpenCL/m07800_a1.cl
index 30b97d567..3d726951e 100644
--- a/OpenCL/m07800_a1.cl
+++ b/OpenCL/m07800_a1.cl
@@ -5,6 +5,9 @@
 
 #define _SAPG_
 
+//incompatible data-dependant code
+//#define NEW_SIMD_CODE
+
 #include "include/constants.h"
 #include "include/kernel_vendor.h"
 
@@ -16,9 +19,7 @@
 #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"
 
 #define GETSHIFTEDINT(a,n) amd_bytealign ((a)[((n)/4)+1], (a)[((n)/4)+0], (n))
 
@@ -206,41 +207,20 @@ __kernel void m07800_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 pw_buf0[4];
+  u32 pw_buf1[4];
 
-  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
    */
@@ -262,108 +242,104 @@ __kernel void m07800_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 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];
-
-    u32 wordr1[4];
+    const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
 
-    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];
+    const u32x pw_len = pw_l_len + pw_r_len;
 
-    u32 wordr2[4];
-
-    wordr2[0] = 0;
-    wordr2[1] = 0;
-    wordr2[2] = 0;
-    wordr2[3] = 0;
-
-    u32 wordr3[4];
+    /**
+     * concat password candidate
+     */
 
-    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);
     }
 
+    u32x w0[4];
+    u32x w1[4];
+    u32x w2[4];
+    u32x w3[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];
+    w1[0] = wordl1[0] | wordr1[0];
+    w1[1] = wordl1[1] | wordr1[1];
+    w1[2] = wordl1[2] | wordr1[2];
+    w1[3] = wordl1[3] | wordr1[3];
+    w2[0] = wordl2[0] | wordr2[0];
+    w2[1] = wordl2[1] | wordr2[1];
+    w2[2] = wordl2[2] | wordr2[2];
+    w2[3] = wordl2[3] | wordr2[3];
+    w3[0] = wordl3[0] | wordr3[0];
+    w3[1] = wordl3[1] | wordr3[1];
+    w3[2] = wordl3[2] | wordr3[2];
+    w3[3] = wordl3[3] | wordr3[3];
+
     /**
-     * append salt
+     * SAP
      */
 
     u32 s0[4];
+    u32 s1[4];
+    u32 s2[4];
+    u32 s3[4];
 
     s0[0] = salt_buf[0];
     s0[1] = salt_buf[1];
     s0[2] = salt_buf[2];
     s0[3] = salt_buf[3];
-
-    u32 s1[4];
-
     s1[0] = salt_buf[4];
     s1[1] = salt_buf[5];
     s1[2] = salt_buf[6];
     s1[3] = salt_buf[7];
-
-    u32 s2[4];
-
     s2[0] = 0;
     s2[1] = 0;
     s2[2] = 0;
     s2[3] = 0;
-
-    u32 s3[4];
-
     s3[0] = 0;
     s3[1] = 0;
     s3[2] = 0;
     s3[3] = 0;
 
-    switch_buffer_by_offset_le (s0, s1, s2, s3, pw_len);
-
-    const u32 pw_salt_len = pw_len + salt_len;
-
-    u32 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];
-
-    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];
-
-    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];
+    switch_buffer_by_offset_le_VV (s0, s1, s2, s3, pw_len);
 
-    w3[0] = wordl3[0] | wordr3[0];
-    w3[1] = wordl3[1] | wordr3[1];
-    w3[2] = 0;
-    w3[3] = 0;
+    const u32x pw_salt_len = pw_len + salt_len;
 
     /**
      * sha1
@@ -495,12 +471,7 @@ __kernel void m07800_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }
 
@@ -528,49 +499,20 @@ __kernel void m07800_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 pw_buf0[4];
+  u32 pw_buf1[4];
 
-  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);
-  }
-
-  const u32 search[4] =
-  {
-    digests_buf[digests_offset].digest_buf[DGST_R0],
-    digests_buf[digests_offset].digest_buf[DGST_R1],
-    digests_buf[digests_offset].digest_buf[DGST_R2],
-    digests_buf[digests_offset].digest_buf[DGST_R3]
-  };
-
   /**
    * salt
    */
@@ -589,111 +531,119 @@ __kernel void m07800_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
   const u32 salt_len = salt_bufs[salt_pos].salt_len;
 
   /**
-   * loop
+   * digest
    */
 
-  for (u32 il_pos = 0; il_pos < il_cnt; il_pos++)
+  const u32 search[4] =
   {
-    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];
-
-    u32 wordr1[4];
+    digests_buf[digests_offset].digest_buf[DGST_R0],
+    digests_buf[digests_offset].digest_buf[DGST_R1],
+    digests_buf[digests_offset].digest_buf[DGST_R2],
+    digests_buf[digests_offset].digest_buf[DGST_R3]
+  };
 
-    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];
+  /**
+   * loop
+   */
 
-    u32 wordr2[4];
+  for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
+  {
+    const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
 
-    wordr2[0] = 0;
-    wordr2[1] = 0;
-    wordr2[2] = 0;
-    wordr2[3] = 0;
+    const u32x pw_len = pw_l_len + pw_r_len;
 
-    u32 wordr3[4];
+    /**
+     * concat password candidate
+     */
 
-    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);
     }
 
+    u32x w0[4];
+    u32x w1[4];
+    u32x w2[4];
+    u32x w3[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];
+    w1[0] = wordl1[0] | wordr1[0];
+    w1[1] = wordl1[1] | wordr1[1];
+    w1[2] = wordl1[2] | wordr1[2];
+    w1[3] = wordl1[3] | wordr1[3];
+    w2[0] = wordl2[0] | wordr2[0];
+    w2[1] = wordl2[1] | wordr2[1];
+    w2[2] = wordl2[2] | wordr2[2];
+    w2[3] = wordl2[3] | wordr2[3];
+    w3[0] = wordl3[0] | wordr3[0];
+    w3[1] = wordl3[1] | wordr3[1];
+    w3[2] = wordl3[2] | wordr3[2];
+    w3[3] = wordl3[3] | wordr3[3];
+
     /**
-     * append salt
+     * SAP
      */
 
     u32 s0[4];
+    u32 s1[4];
+    u32 s2[4];
+    u32 s3[4];
 
     s0[0] = salt_buf[0];
     s0[1] = salt_buf[1];
     s0[2] = salt_buf[2];
     s0[3] = salt_buf[3];
-
-    u32 s1[4];
-
     s1[0] = salt_buf[4];
     s1[1] = salt_buf[5];
     s1[2] = salt_buf[6];
     s1[3] = salt_buf[7];
-
-    u32 s2[4];
-
     s2[0] = 0;
     s2[1] = 0;
     s2[2] = 0;
     s2[3] = 0;
-
-    u32 s3[4];
-
     s3[0] = 0;
     s3[1] = 0;
     s3[2] = 0;
     s3[3] = 0;
 
-    switch_buffer_by_offset_le (s0, s1, s2, s3, pw_len);
-
-    const u32 pw_salt_len = pw_len + salt_len;
+    switch_buffer_by_offset_le_VV (s0, s1, s2, s3, pw_len);
 
-    u32 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];
-
-    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];
-
-    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];
-
-    w3[0] = wordl3[0] | wordr3[0];
-    w3[1] = wordl3[1] | wordr3[1];
-    w3[2] = 0;
-    w3[3] = 0;
+    const u32x pw_salt_len = pw_len + salt_len;
 
     /**
      * sha1
@@ -825,12 +775,7 @@ __kernel void m07800_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }
 
diff --git a/OpenCL/m07800_a3.cl b/OpenCL/m07800_a3.cl
index 78eafc2a3..049f02c30 100644
--- a/OpenCL/m07800_a3.cl
+++ b/OpenCL/m07800_a3.cl
@@ -5,6 +5,9 @@
 
 #define _SAPG_
 
+//incompatible data-dependant code
+//#define NEW_SIMD_CODE
+
 #include "include/constants.h"
 #include "include/kernel_vendor.h"
 
@@ -16,9 +19,7 @@
 #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"
 
 #define GETSHIFTEDINT(a,n) amd_bytealign ((a)[((n)/4)+1], (a)[((n)/4)+0], (n))
 
@@ -199,16 +200,6 @@ static void m07800m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
   const u32 gid = get_global_id (0);
   const u32 lid = get_local_id (0);
 
-  w0[0] = swap32 (w0[0]);
-  w0[1] = swap32 (w0[1]);
-  w0[2] = swap32 (w0[2]);
-  w0[3] = swap32 (w0[3]);
-
-  w1[0] = swap32 (w1[0]);
-  w1[1] = swap32 (w1[1]);
-  w1[2] = swap32 (w1[2]);
-  w1[3] = swap32 (w1[3]);
-
   /**
    * salt
    */
@@ -227,36 +218,30 @@ static void m07800m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
   const u32 salt_len = salt_bufs[salt_pos].salt_len;
 
   u32 s0[4];
-
-  s0[0] = salt_buf[0];
-  s0[1] = salt_buf[1];
-  s0[2] = salt_buf[2];
-  s0[3] = salt_buf[3];
-
   u32 s1[4];
-
-  s1[0] = salt_buf[4];
-  s1[1] = salt_buf[5];
-  s1[2] = salt_buf[6];
-  s1[3] = salt_buf[7];
-
   u32 s2[4];
+  u32 s3[4];
 
+  s0[0] = swap32 (salt_buf[0]);
+  s0[1] = swap32 (salt_buf[1]);
+  s0[2] = swap32 (salt_buf[2]);
+  s0[3] = swap32 (salt_buf[3]);
+  s1[0] = swap32 (salt_buf[4]);
+  s1[1] = swap32 (salt_buf[5]);
+  s1[2] = swap32 (salt_buf[6]);
+  s1[3] = swap32 (salt_buf[7]);
   s2[0] = 0;
   s2[1] = 0;
   s2[2] = 0;
   s2[3] = 0;
-
-  u32 s3[4];
-
   s3[0] = 0;
   s3[1] = 0;
   s3[2] = 0;
   s3[3] = 0;
 
-  switch_buffer_by_offset_le (s0, s1, s2, s3, pw_len);
+  switch_buffer_by_offset_be_S (s0, s1, s2, s3, pw_len);
 
-  const u32 pw_salt_len = pw_len + salt_len;
+  const u32x pw_salt_len = pw_len + salt_len;
 
   /**
    * loop
@@ -264,32 +249,34 @@ static void m07800m (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 < il_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
   {
-    const u32 w0r = swap32 (bfs_buf[il_pos].i);
+    const u32x w0r = ix_create_bft (bfs_buf, il_pos);
+
+    const u32x w0lr = w0l | w0r;
 
-    w0[0] = w0l | w0r;
+    w0[0] = w0lr;
 
     /**
-     * sha1
+     * SAP
      */
 
     u32 final[32];
 
-    final[ 0] = swap32 (w0[0] | s0[0]);
-    final[ 1] = swap32 (w0[1] | s0[1]);
-    final[ 2] = swap32 (w0[2] | s0[2]);
-    final[ 3] = swap32 (w0[3] | s0[3]);
-    final[ 4] = swap32 (w1[0] | s1[0]);
-    final[ 5] = swap32 (w1[1] | s1[1]);
-    final[ 6] = swap32 (w1[2] | s1[2]);
-    final[ 7] = swap32 (w1[3] | s1[3]);
-    final[ 8] = swap32 (w2[0] | s2[0]);
-    final[ 9] = swap32 (w2[1] | s2[1]);
-    final[10] = swap32 (w2[2] | s2[2]);
-    final[11] = swap32 (w2[3] | s2[3]);
-    final[12] = swap32 (w3[0] | s3[0]);
-    final[13] = swap32 (w3[1] | s3[1]);
+    final[ 0] = w0[0] | s0[0];
+    final[ 1] = w0[1] | s0[1];
+    final[ 2] = w0[2] | s0[2];
+    final[ 3] = w0[3] | s0[3];
+    final[ 4] = w1[0] | s1[0];
+    final[ 5] = w1[1] | s1[1];
+    final[ 6] = w1[2] | s1[2];
+    final[ 7] = w1[3] | s1[3];
+    final[ 8] = w2[0] | s2[0];
+    final[ 9] = w2[1] | s2[1];
+    final[10] = w2[2] | s2[2];
+    final[11] = w2[3] | s2[3];
+    final[12] = w3[0] | s3[0];
+    final[13] = w3[1] | s3[1];
     final[14] = 0;
     final[15] = pw_salt_len * 8;
 
@@ -337,21 +324,21 @@ static void m07800m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
     digest[3] = SHA1M_D;
     digest[4] = SHA1M_E;
 
-    #pragma unroll
+    #pragma unroll 32
     for (int i = 0; i < 32; i++) final[i] = 0;
 
-    final[0] = w0[0];
-    final[1] = w0[1];
-    final[2] = w0[2];
-    final[3] = w0[3];
-    final[4] = w1[0];
-    final[5] = w1[1];
-    final[6] = w1[2];
-    final[7] = w1[3];
+    final[0] = swap32 (w0[0]);
+    final[1] = swap32 (w0[1]);
+    final[2] = swap32 (w0[2]);
+    final[3] = swap32 (w0[3]);
+    final[4] = swap32 (w1[0]);
+    final[5] = swap32 (w1[1]);
+    final[6] = swap32 (w1[2]);
+    final[7] = swap32 (w1[3]);
 
     u32 final_len = pw_len;
 
-    u32 i;
+    int i;
 
     // append MagicArray
 
@@ -400,12 +387,7 @@ static void m07800m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }
 
@@ -418,16 +400,6 @@ static void m07800s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
   const u32 gid = get_global_id (0);
   const u32 lid = get_local_id (0);
 
-  w0[0] = swap32 (w0[0]);
-  w0[1] = swap32 (w0[1]);
-  w0[2] = swap32 (w0[2]);
-  w0[3] = swap32 (w0[3]);
-
-  w1[0] = swap32 (w1[0]);
-  w1[1] = swap32 (w1[1]);
-  w1[2] = swap32 (w1[2]);
-  w1[3] = swap32 (w1[3]);
-
   /**
    * salt
    */
@@ -446,36 +418,30 @@ static void m07800s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
   const u32 salt_len = salt_bufs[salt_pos].salt_len;
 
   u32 s0[4];
-
-  s0[0] = salt_buf[0];
-  s0[1] = salt_buf[1];
-  s0[2] = salt_buf[2];
-  s0[3] = salt_buf[3];
-
   u32 s1[4];
-
-  s1[0] = salt_buf[4];
-  s1[1] = salt_buf[5];
-  s1[2] = salt_buf[6];
-  s1[3] = salt_buf[7];
-
   u32 s2[4];
+  u32 s3[4];
 
+  s0[0] = swap32 (salt_buf[0]);
+  s0[1] = swap32 (salt_buf[1]);
+  s0[2] = swap32 (salt_buf[2]);
+  s0[3] = swap32 (salt_buf[3]);
+  s1[0] = swap32 (salt_buf[4]);
+  s1[1] = swap32 (salt_buf[5]);
+  s1[2] = swap32 (salt_buf[6]);
+  s1[3] = swap32 (salt_buf[7]);
   s2[0] = 0;
   s2[1] = 0;
   s2[2] = 0;
   s2[3] = 0;
-
-  u32 s3[4];
-
   s3[0] = 0;
   s3[1] = 0;
   s3[2] = 0;
   s3[3] = 0;
 
-  switch_buffer_by_offset_le (s0, s1, s2, s3, pw_len);
+  switch_buffer_by_offset_be_S (s0, s1, s2, s3, pw_len);
 
-  const u32 pw_salt_len = pw_len + salt_len;
+  const u32x pw_salt_len = pw_len + salt_len;
 
   /**
    * digest
@@ -495,32 +461,34 @@ static void m07800s (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 < il_cnt; il_pos++)
+  for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
   {
-    const u32 w0r = swap32 (bfs_buf[il_pos].i);
+    const u32x w0r = ix_create_bft (bfs_buf, il_pos);
+
+    const u32x w0lr = w0l | w0r;
 
-    w0[0] = w0l | w0r;
+    w0[0] = w0lr;
 
     /**
-     * sha1
+     * SAP
      */
 
     u32 final[32];
 
-    final[ 0] = swap32 (w0[0] | s0[0]);
-    final[ 1] = swap32 (w0[1] | s0[1]);
-    final[ 2] = swap32 (w0[2] | s0[2]);
-    final[ 3] = swap32 (w0[3] | s0[3]);
-    final[ 4] = swap32 (w1[0] | s1[0]);
-    final[ 5] = swap32 (w1[1] | s1[1]);
-    final[ 6] = swap32 (w1[2] | s1[2]);
-    final[ 7] = swap32 (w1[3] | s1[3]);
-    final[ 8] = swap32 (w2[0] | s2[0]);
-    final[ 9] = swap32 (w2[1] | s2[1]);
-    final[10] = swap32 (w2[2] | s2[2]);
-    final[11] = swap32 (w2[3] | s2[3]);
-    final[12] = swap32 (w3[0] | s3[0]);
-    final[13] = swap32 (w3[1] | s3[1]);
+    final[ 0] = w0[0] | s0[0];
+    final[ 1] = w0[1] | s0[1];
+    final[ 2] = w0[2] | s0[2];
+    final[ 3] = w0[3] | s0[3];
+    final[ 4] = w1[0] | s1[0];
+    final[ 5] = w1[1] | s1[1];
+    final[ 6] = w1[2] | s1[2];
+    final[ 7] = w1[3] | s1[3];
+    final[ 8] = w2[0] | s2[0];
+    final[ 9] = w2[1] | s2[1];
+    final[10] = w2[2] | s2[2];
+    final[11] = w2[3] | s2[3];
+    final[12] = w3[0] | s3[0];
+    final[13] = w3[1] | s3[1];
     final[14] = 0;
     final[15] = pw_salt_len * 8;
 
@@ -571,18 +539,18 @@ static void m07800s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
     #pragma unroll 32
     for (int i = 0; i < 32; i++) final[i] = 0;
 
-    final[0] = w0[0];
-    final[1] = w0[1];
-    final[2] = w0[2];
-    final[3] = w0[3];
-    final[4] = w1[0];
-    final[5] = w1[1];
-    final[6] = w1[2];
-    final[7] = w1[3];
+    final[0] = swap32 (w0[0]);
+    final[1] = swap32 (w0[1]);
+    final[2] = swap32 (w0[2]);
+    final[3] = swap32 (w0[3]);
+    final[4] = swap32 (w1[0]);
+    final[5] = swap32 (w1[1]);
+    final[6] = swap32 (w1[2]);
+    final[7] = swap32 (w1[3]);
 
     u32 final_len = pw_len;
 
-    u32 i;
+    int i;
 
     // append MagicArray
 
@@ -631,12 +599,7 @@ static void m07800s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le
 
     sha1_transform (&final[off + 0], &final[off + 4], &final[off + 8], &final[off + 12], 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]);
   }
 }