From 9f9ed78ca7d05b3b68e4d7dceb1db7339cc9e743 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Wed, 4 Mar 2020 15:19:55 +0100
Subject: [PATCH 01/11] Fix -m 7800 and -m 7801 on CPU

---
 OpenCL/m07800_a0-optimized.cl | 4 ++--
 OpenCL/m07800_a1-optimized.cl | 4 ++--
 OpenCL/m07800_a3-optimized.cl | 8 ++++----
 OpenCL/m07801_a0-optimized.cl | 4 ++--
 OpenCL/m07801_a1-optimized.cl | 4 ++--
 OpenCL/m07801_a3-optimized.cl | 8 ++++----
 6 files changed, 16 insertions(+), 16 deletions(-)

diff --git a/OpenCL/m07800_a0-optimized.cl b/OpenCL/m07800_a0-optimized.cl
index 92f46097c..8f921f42f 100644
--- a/OpenCL/m07800_a0-optimized.cl
+++ b/OpenCL/m07800_a0-optimized.cl
@@ -152,7 +152,7 @@ KERNEL_FQ void m07800_m04 (KERN_ATTR_RULES ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -457,7 +457,7 @@ KERNEL_FQ void m07800_s04 (KERN_ATTR_RULES ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
diff --git a/OpenCL/m07800_a1-optimized.cl b/OpenCL/m07800_a1-optimized.cl
index 1449476e8..5b43a174d 100644
--- a/OpenCL/m07800_a1-optimized.cl
+++ b/OpenCL/m07800_a1-optimized.cl
@@ -150,7 +150,7 @@ KERNEL_FQ void m07800_m04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -519,7 +519,7 @@ KERNEL_FQ void m07800_s04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
diff --git a/OpenCL/m07800_a3-optimized.cl b/OpenCL/m07800_a3-optimized.cl
index fd58ed661..a5370badb 100644
--- a/OpenCL/m07800_a3-optimized.cl
+++ b/OpenCL/m07800_a3-optimized.cl
@@ -626,7 +626,7 @@ KERNEL_FQ void m07800_m04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -711,7 +711,7 @@ KERNEL_FQ void m07800_m08 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -800,7 +800,7 @@ KERNEL_FQ void m07800_s04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -885,7 +885,7 @@ KERNEL_FQ void m07800_s08 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
diff --git a/OpenCL/m07801_a0-optimized.cl b/OpenCL/m07801_a0-optimized.cl
index c464d83dc..ba695ee76 100644
--- a/OpenCL/m07801_a0-optimized.cl
+++ b/OpenCL/m07801_a0-optimized.cl
@@ -152,7 +152,7 @@ KERNEL_FQ void m07801_m04 (KERN_ATTR_RULES ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -457,7 +457,7 @@ KERNEL_FQ void m07801_s04 (KERN_ATTR_RULES ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
diff --git a/OpenCL/m07801_a1-optimized.cl b/OpenCL/m07801_a1-optimized.cl
index 4cc78528f..96c5a9e73 100644
--- a/OpenCL/m07801_a1-optimized.cl
+++ b/OpenCL/m07801_a1-optimized.cl
@@ -150,7 +150,7 @@ KERNEL_FQ void m07801_m04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -519,7 +519,7 @@ KERNEL_FQ void m07801_s04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
diff --git a/OpenCL/m07801_a3-optimized.cl b/OpenCL/m07801_a3-optimized.cl
index f1a03c6fd..7c3d987f6 100644
--- a/OpenCL/m07801_a3-optimized.cl
+++ b/OpenCL/m07801_a3-optimized.cl
@@ -626,7 +626,7 @@ KERNEL_FQ void m07801_m04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -711,7 +711,7 @@ KERNEL_FQ void m07801_m08 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -800,7 +800,7 @@ KERNEL_FQ void m07801_s04 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 
@@ -885,7 +885,7 @@ KERNEL_FQ void m07801_s08 (KERN_ATTR_BASIC ())
 
   #else
 
-  CONSTANT_AS u32a *s_theMagicArray = theMagicArray;
+  CONSTANT_AS u32a (*s_theMagicArray)[16] = theMagicArray;
 
   #endif
 

From a4db1a0abd47fb27d50b2ad6b5c86a3eefcfc848 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Thu, 5 Mar 2020 12:43:48 +0100
Subject: [PATCH 02/11] Remove unused OPTS_TYPE_ST_HASH_MD5

---
 include/types.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/include/types.h b/include/types.h
index fabf9f063..a0f800ed2 100644
--- a/include/types.h
+++ b/include/types.h
@@ -407,7 +407,6 @@ typedef enum opts_type
   OPTS_TYPE_ST_ADDBITS15      = (1ULL << 24),
   OPTS_TYPE_ST_HEX            = (1ULL << 25),
   OPTS_TYPE_ST_BASE64         = (1ULL << 26),
-  OPTS_TYPE_ST_HASH_MD5       = (1ULL << 27),
   OPTS_TYPE_HASH_COPY         = (1ULL << 28),
   OPTS_TYPE_HASH_SPLIT        = (1ULL << 29),
   OPTS_TYPE_LOOP_EXTENDED     = (1ULL << 30), // a kernel which is called each time normal _loop kernel finished.

From f8f5e1cc931b3fcc051b4fb8f047e03394f8a003 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 08:53:43 +0100
Subject: [PATCH 03/11] Remove unused OPTS_TYPE_PREFERED_THREAD

---
 include/types.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/include/types.h b/include/types.h
index a0f800ed2..a17f9d10f 100644
--- a/include/types.h
+++ b/include/types.h
@@ -420,7 +420,6 @@ typedef enum opts_type
   OPTS_TYPE_AUX3              = (1ULL << 37),
   OPTS_TYPE_AUX4              = (1ULL << 38),
   OPTS_TYPE_BINARY_HASHFILE   = (1ULL << 39),
-  OPTS_TYPE_PREFERED_THREAD   = (1ULL << 40), // some algorithms (complicated ones with many branches) benefit from this
   OPTS_TYPE_PT_ADD06          = (1ULL << 41),
   OPTS_TYPE_KEYBOARD_MAPPING  = (1ULL << 42),
   OPTS_TYPE_DEEP_COMP_KERNEL  = (1ULL << 43), // if we have to iterate through each hash inside the comp kernel, for example if each hash has to be decrypted separately

From 7f55f69d7fe3bd8efaf7b8244be16066f4034167 Mon Sep 17 00:00:00 2001
From: philsmd <921533+philsmd@users.noreply.github.com>
Date: Fri, 6 Mar 2020 11:49:22 +0100
Subject: [PATCH 04/11] solve telegram format conflict with jtr

---
 src/modules/module_22301.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/modules/module_22301.c b/src/modules/module_22301.c
index 9f87465c7..cda57a6e7 100644
--- a/src/modules/module_22301.c
+++ b/src/modules/module_22301.c
@@ -27,7 +27,7 @@ static const u32   OPTI_TYPE      = OPTI_TYPE_ZERO_BYTE
 static const u64   OPTS_TYPE      = OPTS_TYPE_PT_GENERATE_BE;
 static const u32   SALT_TYPE      = SALT_TYPE_GENERIC;
 static const char *ST_PASS        = "hashcat";
-static const char *ST_HASH        = "$telegram$1*518c001aeb3b4ae96c6173be4cebe60a85f67b1e087b045935849e2f815b5e41*25184098058621950709328221838128";
+static const char *ST_HASH        = "$telegram$0*518c001aeb3b4ae96c6173be4cebe60a85f67b1e087b045935849e2f815b5e41*25184098058621950709328221838128";
 
 u32         module_attack_exec    (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return ATTACK_EXEC;     }
 u32         module_dgst_pos0      (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS0;       }
@@ -84,7 +84,7 @@ int module_hash_decode (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSE
 
   const u8 *version_pos = token.buf[1];
 
-  if (version_pos[0] != '1') return (PARSER_SALT_VALUE);
+  if (version_pos[0] != '0') return (PARSER_SALT_VALUE);
 
   const u8 *hash_pos = token.buf[2];
 
@@ -162,7 +162,7 @@ int module_hash_encode (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSE
 
   const int line_len = snprintf (line_buf, line_size, "%s%i*%08x%08x%08x%08x%08x%08x%08x%08x*%08x%08x%08x%08x",
     SIGNATURE_TELEGRAM,
-    1,
+    0,
     tmp[0],
     tmp[1],
     tmp[2],

From eb46c829983849c81df2c9ebfa4d35d2683ca266 Mon Sep 17 00:00:00 2001
From: philsmd <921533+philsmd@users.noreply.github.com>
Date: Fri, 6 Mar 2020 11:50:16 +0100
Subject: [PATCH 05/11] tests: solve telegram format conflict with jtr

---
 tools/test_modules/m22301.pm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/test_modules/m22301.pm b/tools/test_modules/m22301.pm
index 6a3a32e79..6c708d513 100644
--- a/tools/test_modules/m22301.pm
+++ b/tools/test_modules/m22301.pm
@@ -21,7 +21,7 @@ sub module_generate_hash
 
   my $digest = sha256_hex ($salt_bin . $word . $salt_bin);
 
-  my $hash = sprintf ("\$telegram\$1*%s*%s", $digest, $salt);
+  my $hash = sprintf ("\$telegram\$0*%s*%s", $digest, $salt);
 
   return $hash;
 }
@@ -40,7 +40,7 @@ sub module_verify_hash
 
   my $version = substr ($data[0], 10);
 
-  return unless ($version eq "1");
+  return unless ($version eq "0");
 
   my $digest = $data[1];
   my $salt   = $data[2];

From fa4b521d484d7ced67376ade7f1b97ab93115ac7 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 13:31:32 +0100
Subject: [PATCH 06/11] Add unpack_v8x_from_v32 for vector datatypes, update -m
 200

---
 OpenCL/inc_common.cl          | 188 +++++++++++++++++++++++
 OpenCL/inc_common.h           |   5 +
 OpenCL/m00200_a0-optimized.cl |  40 ++---
 OpenCL/m00200_a1-optimized.cl |  42 +++---
 OpenCL/m00200_a3-optimized.cl | 272 ++++++++++++++--------------------
 5 files changed, 347 insertions(+), 200 deletions(-)

diff --git a/OpenCL/inc_common.cl b/OpenCL/inc_common.cl
index 6a7373867..407a24ef6 100644
--- a/OpenCL/inc_common.cl
+++ b/OpenCL/inc_common.cl
@@ -490,6 +490,194 @@ DECLSPEC u64 v64_from_v32ab_S (const u32 v32a, const u32 v32b)
 
 // unpack function are similar, but always return u32
 
+DECLSPEC u32x unpack_v8a_from_v32 (const u32x v32)
+{
+  u32x r = 0;
+
+  #if   defined IS_NV  && HAS_BFE  == 1
+
+  #if VECT_SIZE == 1
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r) : "r"(v32));
+  #endif
+
+  #if VECT_SIZE >= 2
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s0) : "r"(v32.s0));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s1) : "r"(v32.s1));
+  #endif
+
+  #if VECT_SIZE >= 4
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s2) : "r"(v32.s2));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s3) : "r"(v32.s3));
+  #endif
+
+  #if VECT_SIZE >= 8
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s4) : "r"(v32.s4));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s5) : "r"(v32.s5));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s6) : "r"(v32.s6));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s7) : "r"(v32.s7));
+  #endif
+
+  #if VECT_SIZE >= 16
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s8) : "r"(v32.s8));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.s9) : "r"(v32.s9));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.sa) : "r"(v32.sa));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.sb) : "r"(v32.sb));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.sc) : "r"(v32.sc));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.sd) : "r"(v32.sd));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.se) : "r"(v32.se));
+  asm volatile ("bfe.u32 %0, %1,  0, 8;" : "=r"(r.sf) : "r"(v32.sf));
+  #endif
+
+  //#elif defined IS_AMD && HAS_VBFE == 1
+  //__asm__ __volatile__ ("V_BFE_U32 %0, %1, 0, 8;" : "=v"(r) : "v"(v32));
+  #else
+  r = (v32 >> 0) & 0xff;
+  #endif
+
+  return r;
+}
+
+DECLSPEC u32x unpack_v8b_from_v32 (const u32x v32)
+{
+  u32x r = 0;
+
+  #if   defined IS_NV  && HAS_BFE  == 1
+
+  #if VECT_SIZE == 1
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r) : "r"(v32));
+  #endif
+
+  #if VECT_SIZE >= 2
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s0) : "r"(v32.s0));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s1) : "r"(v32.s1));
+  #endif
+
+  #if VECT_SIZE >= 4
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s2) : "r"(v32.s2));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s3) : "r"(v32.s3));
+  #endif
+
+  #if VECT_SIZE >= 8
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s4) : "r"(v32.s4));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s5) : "r"(v32.s5));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s6) : "r"(v32.s6));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s7) : "r"(v32.s7));
+  #endif
+
+  #if VECT_SIZE >= 16
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s8) : "r"(v32.s8));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.s9) : "r"(v32.s9));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.sa) : "r"(v32.sa));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.sb) : "r"(v32.sb));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.sc) : "r"(v32.sc));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.sd) : "r"(v32.sd));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.se) : "r"(v32.se));
+  asm volatile ("bfe.u32 %0, %1,  8, 8;" : "=r"(r.sf) : "r"(v32.sf));
+  #endif
+
+  //#elif defined IS_AMD && HAS_VBFE == 1
+  //__asm__ __volatile__ ("V_BFE_U32 %0, %1, 8, 8;" : "=v"(r) : "v"(v32));
+  #else
+  r = (v32 >> 8) & 0xff;
+  #endif
+
+  return r;
+}
+
+DECLSPEC u32x unpack_v8c_from_v32 (const u32x v32)
+{
+  u32x r = 0;
+
+  #if   defined IS_NV  && HAS_BFE  == 1
+
+  #if VECT_SIZE == 1
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r) : "r"(v32));
+  #endif
+
+  #if VECT_SIZE >= 2
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s0) : "r"(v32.s0));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s1) : "r"(v32.s1));
+  #endif
+
+  #if VECT_SIZE >= 4
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s2) : "r"(v32.s2));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s3) : "r"(v32.s3));
+  #endif
+
+  #if VECT_SIZE >= 8
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s4) : "r"(v32.s4));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s5) : "r"(v32.s5));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s6) : "r"(v32.s6));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s7) : "r"(v32.s7));
+  #endif
+
+  #if VECT_SIZE >= 16
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s8) : "r"(v32.s8));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.s9) : "r"(v32.s9));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.sa) : "r"(v32.sa));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.sb) : "r"(v32.sb));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.sc) : "r"(v32.sc));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.sd) : "r"(v32.sd));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.se) : "r"(v32.se));
+  asm volatile ("bfe.u32 %0, %1, 16, 8;" : "=r"(r.sf) : "r"(v32.sf));
+  #endif
+
+  //#elif defined IS_AMD && HAS_VBFE == 1
+  //__asm__ __volatile__ ("V_BFE_U32 %0, %1, 16, 8;" : "=v"(r) : "v"(v32));
+  #else
+  r = (v32 >> 16) & 0xff;
+  #endif
+
+  return r;
+}
+
+DECLSPEC u32x unpack_v8d_from_v32 (const u32x v32)
+{
+  u32x r = 0;
+
+  #if   defined IS_NV  && HAS_BFE  == 1
+
+  #if VECT_SIZE == 1
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r) : "r"(v32));
+  #endif
+
+  #if VECT_SIZE >= 2
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s0) : "r"(v32.s0));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s1) : "r"(v32.s1));
+  #endif
+
+  #if VECT_SIZE >= 4
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s2) : "r"(v32.s2));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s3) : "r"(v32.s3));
+  #endif
+
+  #if VECT_SIZE >= 8
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s4) : "r"(v32.s4));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s5) : "r"(v32.s5));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s6) : "r"(v32.s6));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s7) : "r"(v32.s7));
+  #endif
+
+  #if VECT_SIZE >= 16
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s8) : "r"(v32.s8));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.s9) : "r"(v32.s9));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.sa) : "r"(v32.sa));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.sb) : "r"(v32.sb));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.sc) : "r"(v32.sc));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.sd) : "r"(v32.sd));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.se) : "r"(v32.se));
+  asm volatile ("bfe.u32 %0, %1, 24, 8;" : "=r"(r.sf) : "r"(v32.sf));
+  #endif
+
+  //#elif defined IS_AMD && HAS_VBFE == 1
+  //__asm__ __volatile__ ("V_BFE_U32 %0, %1, 24, 8;" : "=v"(r) : "v"(v32));
+  #else
+  r = (v32 >> 24) & 0xff;
+  #endif
+
+  return r;
+}
+
 DECLSPEC u32 unpack_v8a_from_v32_S (const u32 v32)
 {
   u32 r = 0;
diff --git a/OpenCL/inc_common.h b/OpenCL/inc_common.h
index 8715ae75e..07137297b 100644
--- a/OpenCL/inc_common.h
+++ b/OpenCL/inc_common.h
@@ -171,6 +171,11 @@ DECLSPEC u64 v64_from_v32ab_S (const u32 v32a, const u32 v32b);
 
 // inline asm packing
 
+DECLSPEC u32x unpack_v8a_from_v32 (const u32x v32);
+DECLSPEC u32x unpack_v8b_from_v32 (const u32x v32);
+DECLSPEC u32x unpack_v8c_from_v32 (const u32x v32);
+DECLSPEC u32x unpack_v8d_from_v32 (const u32x v32);
+
 DECLSPEC u32 unpack_v8a_from_v32_S (const u32 v32);
 DECLSPEC u32 unpack_v8b_from_v32_S (const u32 v32);
 DECLSPEC u32 unpack_v8c_from_v32_S (const u32 v32);
diff --git a/OpenCL/m00200_a0-optimized.cl b/OpenCL/m00200_a0-optimized.cl
index c5d334f36..ab9071cb0 100644
--- a/OpenCL/m00200_a0-optimized.cl
+++ b/OpenCL/m00200_a0-optimized.cl
@@ -99,10 +99,10 @@ KERNEL_FQ void m00200_m04 (KERN_ATTR_RULES ())
     {
       const u32x wj = w_t[j];
 
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
-      ROUND ((wj >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
+      ROUND (unpack_v8d_from_v32 (wj));
     }
 
     const u32x wj = w_t[j];
@@ -111,18 +111,18 @@ KERNEL_FQ void m00200_m04 (KERN_ATTR_RULES ())
 
     if (left == 3)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
     }
     else if (left == 2)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
     }
     else if (left == 1)
     {
-      ROUND ((wj >>  0) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
     }
 
     a &= 0x7fffffff;
@@ -237,10 +237,10 @@ KERNEL_FQ void m00200_s04 (KERN_ATTR_RULES ())
     {
       const u32x wj = w_t[j];
 
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
-      ROUND ((wj >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
+      ROUND (unpack_v8d_from_v32 (wj));
     }
 
     const u32x wj = w_t[j];
@@ -249,18 +249,18 @@ KERNEL_FQ void m00200_s04 (KERN_ATTR_RULES ())
 
     if (left == 3)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
     }
     else if (left == 2)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
     }
     else if (left == 1)
     {
-      ROUND ((wj >>  0) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
     }
 
     a &= 0x7fffffff;
diff --git a/OpenCL/m00200_a1-optimized.cl b/OpenCL/m00200_a1-optimized.cl
index 5589a386a..16b5054ab 100644
--- a/OpenCL/m00200_a1-optimized.cl
+++ b/OpenCL/m00200_a1-optimized.cl
@@ -142,6 +142,8 @@ KERNEL_FQ void m00200_m04 (KERN_ATTR_BASIC ())
 
     u32x a = MYSQL323_A;
     u32x b = MYSQL323_B;
+    u32x c = 0;
+    u32x d = 0;
 
     u32x add = 7;
 
@@ -159,10 +161,10 @@ KERNEL_FQ void m00200_m04 (KERN_ATTR_BASIC ())
     {
       const u32x wj = w_t[j];
 
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
-      ROUND ((wj >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
+      ROUND (unpack_v8d_from_v32 (wj));
     }
 
     const u32x wj = w_t[j];
@@ -171,18 +173,18 @@ KERNEL_FQ void m00200_m04 (KERN_ATTR_BASIC ())
 
     if (left == 3)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
     }
     else if (left == 2)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
     }
     else if (left == 1)
     {
-      ROUND ((wj >>  0) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
     }
 
     a &= 0x7fffffff;
@@ -361,10 +363,10 @@ KERNEL_FQ void m00200_s04 (KERN_ATTR_BASIC ())
     {
       const u32x wj = w_t[j];
 
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
-      ROUND ((wj >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
+      ROUND (unpack_v8d_from_v32 (wj));
     }
 
     const u32x wj = w_t[j];
@@ -373,18 +375,18 @@ KERNEL_FQ void m00200_s04 (KERN_ATTR_BASIC ())
 
     if (left == 3)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
-      ROUND ((wj >> 16) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
+      ROUND (unpack_v8c_from_v32 (wj));
     }
     else if (left == 2)
     {
-      ROUND ((wj >>  0) & 0xff);
-      ROUND ((wj >>  8) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
+      ROUND (unpack_v8b_from_v32 (wj));
     }
     else if (left == 1)
     {
-      ROUND ((wj >>  0) & 0xff);
+      ROUND (unpack_v8a_from_v32 (wj));
     }
 
     a &= 0x7fffffff;
diff --git a/OpenCL/m00200_a3-optimized.cl b/OpenCL/m00200_a3-optimized.cl
index 74a1c3234..1a9b72e39 100644
--- a/OpenCL/m00200_a3-optimized.cl
+++ b/OpenCL/m00200_a3-optimized.cl
@@ -42,10 +42,10 @@
     {                                                             \
       const u32 wj = w[j];                                        \
                                                                   \
-      ROUND ((wj >>  0) & 0xff);                                  \
-      ROUND ((wj >>  8) & 0xff);                                  \
-      ROUND ((wj >> 16) & 0xff);                                  \
-      ROUND ((wj >> 24) & 0xff);                                  \
+      ROUND (unpack_v8a_from_v32 (wj));                           \
+      ROUND (unpack_v8b_from_v32 (wj));                           \
+      ROUND (unpack_v8c_from_v32 (wj));                           \
+      ROUND (unpack_v8d_from_v32 (wj));                           \
     }                                                             \
                                                                   \
     const u32 wj = w[j];                                          \
@@ -54,18 +54,18 @@
                                                                   \
     if (left == 3)                                                \
     {                                                             \
-      ROUND ((wj >>  0) & 0xff);                                  \
-      ROUND ((wj >>  8) & 0xff);                                  \
-      ROUND ((wj >> 16) & 0xff);                                  \
+      ROUND (unpack_v8a_from_v32 (wj));                           \
+      ROUND (unpack_v8b_from_v32 (wj));                           \
+      ROUND (unpack_v8c_from_v32 (wj));                           \
     }                                                             \
     else if (left == 2)                                           \
     {                                                             \
-      ROUND ((wj >>  0) & 0xff);                                  \
-      ROUND ((wj >>  8) & 0xff);                                  \
+      ROUND (unpack_v8a_from_v32 (wj));                           \
+      ROUND (unpack_v8b_from_v32 (wj));                           \
     }                                                             \
     else if (left == 1)                                           \
     {                                                             \
-      ROUND ((wj >>  0) & 0xff);                                  \
+      ROUND (unpack_v8a_from_v32 (wj));                           \
     }
 
 #define CODE_POST_M                                               \
@@ -99,141 +99,123 @@ DECLSPEC void m00200m (u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
   const u64 gid = get_global_id (0);
   const u64 lid = get_local_id (0);
 
+  /**
+   * digest
+   */
+
+  const u32 search[4] =
+  {
+    digests_buf[digests_offset].digest_buf[DGST_R0],
+    digests_buf[digests_offset].digest_buf[DGST_R1],
+    0,
+    0
+  };
+
   /**
    * loop
    */
 
   u32 w0l = w[0];
 
+  CODE_PRE;
+
   switch (pw_len)
   {
     case  1:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0));
       break;
 
     case  2:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0));
       break;
 
     case  3:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0));
       break;
 
     case  4:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
       break;
 
     case  5:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1]));
       break;
 
     case  6:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1]));
       break;
 
     case  7:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1]));
       break;
 
     case  8:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
       break;
 
     case  9:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2]));
       break;
 
     case 10:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2]));
       break;
 
     case 11:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2]));
       break;
 
     case 12:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
       break;
 
     case 13:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3]));
       break;
 
     case 14:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3]));
       break;
 
     case 15:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff); ROUND ((w[3] >> 16) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3])); ROUND (unpack_v8c_from_v32 (w[3]));
       break;
 
     case 16:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff); ROUND ((w[3] >> 16) & 0xff); ROUND ((w[3] >> 24) & 0xff);
-      CODE_POST_M;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3])); ROUND (unpack_v8c_from_v32 (w[3])); ROUND (unpack_v8d_from_v32 (w[3]));
       break;
 
     default:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
       CODE_LOOP (pw_len - 4);
-      CODE_POST_M;
       break;
   }
+
+  CODE_POST_M;
 }
 
 DECLSPEC void m00200s (u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
@@ -263,135 +245,105 @@ DECLSPEC void m00200s (u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
 
   u32 w0l = w[0];
 
+  CODE_PRE;
+
   switch (pw_len)
   {
     case  1:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0));
       break;
 
     case  2:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0));
       break;
 
     case  3:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0));
       break;
 
     case  4:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
       break;
 
     case  5:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1]));
       break;
 
     case  6:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1]));
       break;
 
     case  7:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1]));
       break;
 
     case  8:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
       break;
 
     case  9:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2]));
       break;
 
     case 10:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2]));
       break;
 
     case 11:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2]));
       break;
 
     case 12:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
       break;
 
     case 13:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3]));
       break;
 
     case 14:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3]));
       break;
 
     case 15:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff); ROUND ((w[3] >> 16) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3])); ROUND (unpack_v8c_from_v32 (w[3]));
       break;
 
     case 16:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
-      ROUND ((w[1] >>  0) & 0xff); ROUND ((w[1] >>  8) & 0xff); ROUND ((w[1] >> 16) & 0xff); ROUND ((w[1] >> 24) & 0xff);
-      ROUND ((w[2] >>  0) & 0xff); ROUND ((w[2] >>  8) & 0xff); ROUND ((w[2] >> 16) & 0xff); ROUND ((w[2] >> 24) & 0xff);
-      ROUND ((w[3] >>  0) & 0xff); ROUND ((w[3] >>  8) & 0xff); ROUND ((w[3] >> 16) & 0xff); ROUND ((w[3] >> 24) & 0xff);
-      CODE_POST_S;
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
+      ROUND (unpack_v8a_from_v32 (w[1])); ROUND (unpack_v8b_from_v32 (w[1])); ROUND (unpack_v8c_from_v32 (w[1])); ROUND (unpack_v8d_from_v32 (w[1]));
+      ROUND (unpack_v8a_from_v32 (w[2])); ROUND (unpack_v8b_from_v32 (w[2])); ROUND (unpack_v8c_from_v32 (w[2])); ROUND (unpack_v8d_from_v32 (w[2]));
+      ROUND (unpack_v8a_from_v32 (w[3])); ROUND (unpack_v8b_from_v32 (w[3])); ROUND (unpack_v8c_from_v32 (w[3])); ROUND (unpack_v8d_from_v32 (w[3]));
       break;
 
     default:
-      CODE_PRE;
-      ROUND ((w0   >>  0) & 0xff); ROUND ((w0   >>  8) & 0xff); ROUND ((w0   >> 16) & 0xff); ROUND ((w0   >> 24) & 0xff);
+      ROUND (unpack_v8a_from_v32 (  w0)); ROUND (unpack_v8b_from_v32 (  w0)); ROUND (unpack_v8c_from_v32 (  w0)); ROUND (unpack_v8d_from_v32 (  w0));
       CODE_LOOP (pw_len - 4);
-      CODE_POST_S;
       break;
   }
+
+  CODE_POST_S;
 }
 
 KERNEL_FQ void m00200_m04 (KERN_ATTR_VECTOR ())

From fd06e407cfc362fa20549cc1ab063a0e020dc48b Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 13:44:07 +0100
Subject: [PATCH 07/11] Remove #undef _unroll because _unroll is no longer the
 default

---
 OpenCL/m01500_a3-pure.cl | 1 -
 OpenCL/m03000_a3-pure.cl | 1 -
 OpenCL/m14000_a3-pure.cl | 1 -
 3 files changed, 3 deletions(-)

diff --git a/OpenCL/m01500_a3-pure.cl b/OpenCL/m01500_a3-pure.cl
index 6c45f125b..a751c55fe 100644
--- a/OpenCL/m01500_a3-pure.cl
+++ b/OpenCL/m01500_a3-pure.cl
@@ -16,7 +16,6 @@
 #define COMPARE_M "inc_comp_multi_bs.cl"
 
 #ifdef IS_NV
-#undef _unroll
 #define KXX_DECL
 #endif
 
diff --git a/OpenCL/m03000_a3-pure.cl b/OpenCL/m03000_a3-pure.cl
index 67f29aa4e..27b5c5630 100644
--- a/OpenCL/m03000_a3-pure.cl
+++ b/OpenCL/m03000_a3-pure.cl
@@ -16,7 +16,6 @@
 #define COMPARE_M "inc_comp_multi_bs.cl"
 
 #ifdef IS_NV
-#undef _unroll
 #define KXX_DECL
 #endif
 
diff --git a/OpenCL/m14000_a3-pure.cl b/OpenCL/m14000_a3-pure.cl
index 253a6ee86..5f3abca84 100644
--- a/OpenCL/m14000_a3-pure.cl
+++ b/OpenCL/m14000_a3-pure.cl
@@ -16,7 +16,6 @@
 #define COMPARE_M "inc_comp_multi_bs.cl"
 
 #ifdef IS_NV
-#undef _unroll
 #define KXX_DECL
 #endif
 

From 2fdb7ded8b5ec6e1ede4c602a42626a4b083f0a3 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 14:09:38 +0100
Subject: [PATCH 08/11] Update self-test hash for -m 1500

---
 src/modules/module_01500.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/modules/module_01500.c b/src/modules/module_01500.c
index d2b325339..77ab912d2 100644
--- a/src/modules/module_01500.c
+++ b/src/modules/module_01500.c
@@ -25,7 +25,7 @@ static const u64   OPTS_TYPE      = OPTS_TYPE_PT_GENERATE_LE
                                   | OPTS_TYPE_TM_KERNEL;
 static const u32   SALT_TYPE      = SALT_TYPE_EMBEDDED;
 static const char *ST_PASS        = NULL; // the self-test can't work because the salt is not part of the code at compile-time
-static const char *ST_HASH        = "8133vc.5rieNk";
+static const char *ST_HASH        = "24leDr0hHfb3A";
 
 u32         module_attack_exec    (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return ATTACK_EXEC;     }
 u32         module_dgst_pos0      (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS0;       }

From af1e5f111aa0164a36ae4bc56c427a63054613f0 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 14:23:44 +0100
Subject: [PATCH 09/11] Add missing entry in changes.txt

---
 docs/changes.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/changes.txt b/docs/changes.txt
index dfea6fb71..34f830347 100644
--- a/docs/changes.txt
+++ b/docs/changes.txt
@@ -81,6 +81,7 @@
 - Fixed cracking of Cisco-PIX and Cisco-ASA MD5 passwords in mask-attack mode if mask > length 16
 - Fixed cracking of Electrum Wallet Salt-Type 2 hashes
 - Fixed cracking of NetNTLMv1 passwords in mask-attack mode if mask > length 16 (optimized kernels only)
+- Fixed cracking of VeraCrypt Streebog-512 hashes (CPU only)
 - Fixed cracking raw Streebog-HMAC 256 and 512 hashes with password of length >= 64
 - Fixed cracking raw Whirlpool hashes cracking with password of length >= 32
 - Fixed incorrect progress-only result in a special race condition

From 5b58cba12eec0dcb8a7e66a0ed7c15763184f18d Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 14:34:41 +0100
Subject: [PATCH 10/11] Use _unroll in -m 5500 for CPU

---
 src/modules/module_05500.c | 21 ++++++++++++++++++++-
 1 file changed, 20 insertions(+), 1 deletion(-)

diff --git a/src/modules/module_05500.c b/src/modules/module_05500.c
index 6a57ee25b..1ebabe097 100644
--- a/src/modules/module_05500.c
+++ b/src/modules/module_05500.c
@@ -79,6 +79,25 @@ static void transform_netntlmv1_key (const u8 *nthash, u8 *key)
   key[7] |= 0x01;
 }
 
+char *module_jit_build_options (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra, MAYBE_UNUSED const hashes_t *hashes, MAYBE_UNUSED const hc_device_param_t *device_param)
+{
+  char *jit_build_options = NULL;
+
+  // Extra treatment for Apple systems
+  if (device_param->opencl_platform_vendor_id == VENDOR_ID_APPLE)
+  {
+    return jit_build_options;
+  }
+
+  // Intel CPU
+  if ((device_param->opencl_device_vendor_id == VENDOR_ID_INTEL_SDK) && (device_param->opencl_device_type & CL_DEVICE_TYPE_CPU))
+  {
+    hc_asprintf (&jit_build_options, "-D _unroll");
+  }
+
+  return jit_build_options;
+}
+
 u64 module_esalt_size (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra)
 {
   const u64 esalt_size = (const u64) sizeof (netntlm_t);
@@ -424,7 +443,7 @@ void module_init (module_ctx_t *module_ctx)
   module_ctx->module_hook23                   = MODULE_DEFAULT;
   module_ctx->module_hook_salt_size           = MODULE_DEFAULT;
   module_ctx->module_hook_size                = MODULE_DEFAULT;
-  module_ctx->module_jit_build_options        = MODULE_DEFAULT;
+  module_ctx->module_jit_build_options        = module_jit_build_options;
   module_ctx->module_jit_cache_disable        = MODULE_DEFAULT;
   module_ctx->module_kernel_accel_max         = MODULE_DEFAULT;
   module_ctx->module_kernel_accel_min         = MODULE_DEFAULT;

From b1d5f92c2d98a22f5cd156335a5d4ad7b4069db0 Mon Sep 17 00:00:00 2001
From: Jens Steube <jens.steube@gmail.com>
Date: Fri, 6 Mar 2020 15:48:01 +0100
Subject: [PATCH 11/11] Do not use __local keyword in -m 5500 for devices
 without real shared memory

---
 OpenCL/m05500_a0-optimized.cl | 22 +++++++++++--
 OpenCL/m05500_a0-pure.cl      | 22 +++++++++++--
 OpenCL/m05500_a1-optimized.cl | 22 +++++++++++--
 OpenCL/m05500_a1-pure.cl      | 22 +++++++++++--
 OpenCL/m05500_a3-optimized.cl | 62 ++++++++++++++++++++++++++++++++---
 OpenCL/m05500_a3-pure.cl      | 22 +++++++++++--
 6 files changed, 158 insertions(+), 14 deletions(-)

diff --git a/OpenCL/m05500_a0-optimized.cl b/OpenCL/m05500_a0-optimized.cl
index 6a5c8451c..03dfe4f7a 100644
--- a/OpenCL/m05500_a0-optimized.cl
+++ b/OpenCL/m05500_a0-optimized.cl
@@ -356,7 +356,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32x r = data[0];
   u32x l = data[1];
@@ -398,7 +398,7 @@ DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCA
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32x tt;
 
@@ -516,6 +516,8 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_RULES ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -542,6 +544,13 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_RULES ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -729,6 +738,8 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_RULES ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -755,6 +766,13 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_RULES ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
diff --git a/OpenCL/m05500_a0-pure.cl b/OpenCL/m05500_a0-pure.cl
index 7681eea15..ac9c30f35 100644
--- a/OpenCL/m05500_a0-pure.cl
+++ b/OpenCL/m05500_a0-pure.cl
@@ -356,7 +356,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32 r = data[0];
   u32 l = data[1];
@@ -398,7 +398,7 @@ DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, LOCAL_AS
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32 c, u32 d, u32 *Kc, u32 *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32 c, u32 d, u32 *Kc, u32 *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32 tt;
 
@@ -516,6 +516,8 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_RULES ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -542,6 +544,13 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_RULES ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -640,6 +649,8 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_RULES ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -666,6 +677,13 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_RULES ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
diff --git a/OpenCL/m05500_a1-optimized.cl b/OpenCL/m05500_a1-optimized.cl
index d4ae2526a..39a7ed212 100644
--- a/OpenCL/m05500_a1-optimized.cl
+++ b/OpenCL/m05500_a1-optimized.cl
@@ -354,7 +354,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32x r = data[0];
   u32x l = data[1];
@@ -396,7 +396,7 @@ DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCA
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32x tt;
 
@@ -514,6 +514,8 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_BASIC ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -540,6 +542,13 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_BASIC ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -780,6 +789,8 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_BASIC ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -806,6 +817,13 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_BASIC ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
diff --git a/OpenCL/m05500_a1-pure.cl b/OpenCL/m05500_a1-pure.cl
index 28b5a627c..c53e12357 100644
--- a/OpenCL/m05500_a1-pure.cl
+++ b/OpenCL/m05500_a1-pure.cl
@@ -354,7 +354,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32 r = data[0];
   u32 l = data[1];
@@ -396,7 +396,7 @@ DECLSPEC void _des_crypt_encrypt (u32 *iv, u32 *data, u32 *Kc, u32 *Kd, LOCAL_AS
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32 c, u32 d, u32 *Kc, u32 *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32 c, u32 d, u32 *Kc, u32 *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32 tt;
 
@@ -514,6 +514,8 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_BASIC ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -540,6 +542,13 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_BASIC ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -636,6 +645,8 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_BASIC ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -662,6 +673,13 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_BASIC ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
diff --git a/OpenCL/m05500_a3-optimized.cl b/OpenCL/m05500_a3-optimized.cl
index 90a3c3b14..a52b1cc7c 100644
--- a/OpenCL/m05500_a3-optimized.cl
+++ b/OpenCL/m05500_a3-optimized.cl
@@ -354,7 +354,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32x r = data[0];
   u32x l = data[1];
@@ -396,7 +396,7 @@ DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCA
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32x tt;
 
@@ -500,7 +500,7 @@ DECLSPEC void transform_netntlmv1_key (const u32x w0, const u32x w1, u32x *out)
          | ((k[7] & 0xff) << 24);
 }
 
-DECLSPEC void m05500m (LOCAL_AS u32 (*s_SPtrans)[64], LOCAL_AS u32 (*s_skb)[64], u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
+DECLSPEC void m05500m (SHM_TYPE u32 (*s_SPtrans)[64], SHM_TYPE u32 (*s_skb)[64], u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
 {
   /**
    * modifier
@@ -657,7 +657,7 @@ DECLSPEC void m05500m (LOCAL_AS u32 (*s_SPtrans)[64], LOCAL_AS u32 (*s_skb)[64],
   }
 }
 
-DECLSPEC void m05500s (LOCAL_AS u32 (*s_SPtrans)[64], LOCAL_AS u32 (*s_skb)[64], u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
+DECLSPEC void m05500s (SHM_TYPE u32 (*s_SPtrans)[64], SHM_TYPE u32 (*s_skb)[64], u32 *w, const u32 pw_len, KERN_ATTR_VECTOR ())
 {
   /**
    * modifier
@@ -847,6 +847,8 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -873,6 +875,13 @@ KERNEL_FQ void m05500_m04 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -921,6 +930,8 @@ KERNEL_FQ void m05500_m08 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -947,6 +958,13 @@ KERNEL_FQ void m05500_m08 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -995,6 +1013,8 @@ KERNEL_FQ void m05500_m16 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -1021,6 +1041,13 @@ KERNEL_FQ void m05500_m16 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -1069,6 +1096,8 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -1095,6 +1124,13 @@ KERNEL_FQ void m05500_s04 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -1143,6 +1179,8 @@ KERNEL_FQ void m05500_s08 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -1169,6 +1207,13 @@ KERNEL_FQ void m05500_s08 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -1217,6 +1262,8 @@ KERNEL_FQ void m05500_s16 (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -1243,6 +1290,13 @@ KERNEL_FQ void m05500_s16 (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
diff --git a/OpenCL/m05500_a3-pure.cl b/OpenCL/m05500_a3-pure.cl
index 25e6392fb..e691330cd 100644
--- a/OpenCL/m05500_a3-pure.cl
+++ b/OpenCL/m05500_a3-pure.cl
@@ -354,7 +354,7 @@ CONSTANT_VK u32a c_skb[8][64] =
 #define BOX(i,n,S) make_u32x ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
 #endif
 
-DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_SPtrans)[64])
+DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_SPtrans)[64])
 {
   u32x r = data[0];
   u32x l = data[1];
@@ -396,7 +396,7 @@ DECLSPEC void _des_crypt_encrypt (u32x *iv, u32x *data, u32x *Kc, u32x *Kd, LOCA
   iv[1] = r;
 }
 
-DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, LOCAL_AS u32 (*s_skb)[64])
+DECLSPEC void _des_crypt_keysetup (u32x c, u32x d, u32x *Kc, u32x *Kd, SHM_TYPE u32 (*s_skb)[64])
 {
   u32x tt;
 
@@ -514,6 +514,8 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -540,6 +542,13 @@ KERNEL_FQ void m05500_mxx (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**
@@ -649,6 +658,8 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_VECTOR ())
    * sbox, kbox
    */
 
+  #ifdef REAL_SHM
+
   LOCAL_VK u32 s_SPtrans[8][64];
   LOCAL_VK u32 s_skb[8][64];
 
@@ -675,6 +686,13 @@ KERNEL_FQ void m05500_sxx (KERN_ATTR_VECTOR ())
 
   SYNC_THREADS ();
 
+  #else
+
+  CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans;
+  CONSTANT_AS u32a (*s_skb)[64]     = c_skb;
+
+  #endif
+
   if (gid >= gid_max) return;
 
   /**