arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2024-11-13 19:28:56 +00:00
Code Issues Releases Wiki Activity
0e68b2af25
hashcat/OpenCL/m00200_a3.cl

631 lines
28 KiB
Common Lisp
Raw Normal View History

Initial commit
2015-12-04 14:47:52 +00:00
/**
* Author......: Jens Steube <jens.steube@gmail.com>
* License.....: MIT
*/
#define _MYSQL323_
Converted _a3 kernels, use SIMD for CPU and GPU
2016-01-23 14:32:31 +00:00
#define NEW_SIMD_CODE
Move files from include/ to OpenCL/ if they are used within kernels Rename includes in OpenCL so that it's easier to recognize them as such
2016-05-25 21:04:26 +00:00
#include "inc_hash_constants.h"
#include "inc_vendor.cl"
Initial commit
2015-12-04 14:47:52 +00:00
#define DGST_R0 0
#define DGST_R1 1
#define DGST_R2 2
#define DGST_R3 3
Move files from include/ to OpenCL/ if they are used within kernels Rename includes in OpenCL so that it's easier to recognize them as such
2016-05-25 21:04:26 +00:00
#include "inc_hash_functions.cl"
#include "inc_types.cl"
#include "inc_common.cl"
#include "inc_simd.cl"
Initial commit
2015-12-04 14:47:52 +00:00
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
#define ROUND(v) \
{ \
a ^= (((a & 0x3f) + add) * (v)) + (a << 8); \
b += (b << 8) ^ a; \
add += v; \
}
#define CODE_PRE \
{ \
for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) \
{ \
const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; \
\
const u32x w0 = w0l | w0r; \
\
u32x a = MYSQL323_A; \
u32x b = MYSQL323_B; \
\
u32x add = 7; \
#define CODE_LOOP(rest) \
\
int i; \
int j; \
\
for (i = 0, j = 1; i <= (int) (rest) - 4; i += 4, j += 1) \
{ \
const u32 wj = w[j]; \
\
ROUND ((wj >> 0) & 0xff); \
ROUND ((wj >> 8) & 0xff); \
ROUND ((wj >> 16) & 0xff); \
ROUND ((wj >> 24) & 0xff); \
} \
\
const u32 wj = w[j]; \
\
const u32 left = (rest) - i; \
\
if (left == 3) \
{ \
ROUND ((wj >> 0) & 0xff); \
ROUND ((wj >> 8) & 0xff); \
ROUND ((wj >> 16) & 0xff); \
} \
else if (left == 2) \
{ \
ROUND ((wj >> 0) & 0xff); \
ROUND ((wj >> 8) & 0xff); \
} \
else if (left == 1) \
{ \
ROUND ((wj >> 0) & 0xff); \
}
#define CODE_POST_M \
\
a &= 0x7fffffff; \
b &= 0x7fffffff; \
\
Fix -m 200 multihash and benchmark
2016-05-03 08:39:09 +00:00
u32x z = 0; \
\
COMPARE_M_SIMD (a, b, z, z); \
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
} \
}
#define CODE_POST_S \
\
a &= 0x7fffffff; \
b &= 0x7fffffff; \
\
Fix -m 200 multihash and benchmark
2016-05-03 08:39:09 +00:00
u32x z = 0; \
\
COMPARE_S_SIMD (a, b, z, z); \
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
} \
}
- Added inline declaration to functions from simd.c, common.c, rp.c and types_ocl.c to increase performance - Dropped static declaration from functions in all kernel to achieve OpenCL 1.1 compatibility - Added -cl-std=CL1.1 to all kernel build options - Created environment variable to inform NVidia OpenCL runtime to not create its own kernel cache - Created environment variable to inform pocl OpenCL runtime to not create its own kernel cache
2016-05-01 21:15:26 +00:00
void m00200m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* modifier
*/
const u32 gid = get_global_id (0);
const u32 lid = get_local_id (0);
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
/**
* loop
*/
Initial commit
2015-12-04 14:47:52 +00:00
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
u32 w0l = w[0];
Initial commit
2015-12-04 14:47:52 +00:00
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
switch (pw_len)
{
case 1:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff);
CODE_POST_M;
break;
case 2:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff);
CODE_POST_M;
break;
case 3:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff);
CODE_POST_M;
break;
case 4:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff); ROUND ((w0 >> 24) & 0xff);
CODE_POST_M;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
break;
default:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff); ROUND ((w0 >> 24) & 0xff);
CODE_LOOP (pw_len - 4);
CODE_POST_M;
break;
Initial commit
2015-12-04 14:47:52 +00:00
}
}
- Added inline declaration to functions from simd.c, common.c, rp.c and types_ocl.c to increase performance - Dropped static declaration from functions in all kernel to achieve OpenCL 1.1 compatibility - Added -cl-std=CL1.1 to all kernel build options - Created environment variable to inform NVidia OpenCL runtime to not create its own kernel cache - Created environment variable to inform pocl OpenCL runtime to not create its own kernel cache
2016-05-01 21:15:26 +00:00
void m00200s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* modifier
*/
const u32 gid = get_global_id (0);
const u32 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],
Fix -m 200 multihash and benchmark
2016-05-03 08:39:09 +00:00
0,
0
Initial commit
2015-12-04 14:47:52 +00:00
};
/**
* loop
*/
- Dropped all vector code since new GPU's are all scalar, makes the code much easier - Some performance on low-end GPU may drop because of that, but only for a few hash-modes - Dropped scalar code (aka warp) since we do not have any vector datatypes anymore - Renamed C++ overloading functions memcat32_9 -> memcat_c32_w4x4_a3x4 - Still need to fix kernels to new function names, needs to be done manually - Temperature Management needs to be rewritten partially because of conflicting datatypes names - Added code to create different codepaths for NV on AMD in runtime in host (see data.vendor_id) - Added code to create different codepaths for NV on AMD in runtime in kernels (see IS_NV and IS_AMD) - First tests working for -m 0, for example - Great performance increases in general for NV so far - Tested amp_* and markov_* kernel - Migrated special NV optimizations for rule processor
2015-12-15 11:04:22 +00:00
u32 w0l = w[0];
Initial commit
2015-12-04 14:47:52 +00:00
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
switch (pw_len)
Initial commit
2015-12-04 14:47:52 +00:00
{
Fix -m 200 cracking performance
2016-04-23 18:52:16 +00:00
case 1:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff);
CODE_POST_S;
break;
case 2:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff);
CODE_POST_S;
break;
case 3:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff);
CODE_POST_S;
break;
case 4:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff); ROUND ((w0 >> 24) & 0xff);
CODE_POST_S;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
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;
break;
default:
CODE_PRE;
ROUND ((w0 >> 0) & 0xff); ROUND ((w0 >> 8) & 0xff); ROUND ((w0 >> 16) & 0xff); ROUND ((w0 >> 24) & 0xff);
CODE_LOOP (pw_len - 4);
CODE_POST_S;
break;
Initial commit
2015-12-04 14:47:52 +00:00
}
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = 0;
w[ 5] = 0;
w[ 6] = 0;
w[ 7] = 0;
w[ 8] = 0;
w[ 9] = 0;
w[10] = 0;
w[11] = 0;
w[12] = 0;
w[13] = 0;
w[14] = 0;
w[15] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200m (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_m08 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = pws[gid].i[ 4];
w[ 5] = pws[gid].i[ 5];
w[ 6] = pws[gid].i[ 6];
w[ 7] = pws[gid].i[ 7];
w[ 8] = 0;
w[ 9] = 0;
w[10] = 0;
w[11] = 0;
w[12] = 0;
w[13] = 0;
w[14] = 0;
w[15] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200m (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_m16 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = pws[gid].i[ 4];
w[ 5] = pws[gid].i[ 5];
w[ 6] = pws[gid].i[ 6];
w[ 7] = pws[gid].i[ 7];
w[ 8] = pws[gid].i[ 8];
w[ 9] = pws[gid].i[ 9];
w[10] = pws[gid].i[10];
w[11] = pws[gid].i[11];
w[12] = pws[gid].i[12];
w[13] = pws[gid].i[13];
w[14] = pws[gid].i[14];
w[15] = pws[gid].i[15];
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200m (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = 0;
w[ 5] = 0;
w[ 6] = 0;
w[ 7] = 0;
w[ 8] = 0;
w[ 9] = 0;
w[10] = 0;
w[11] = 0;
w[12] = 0;
w[13] = 0;
w[14] = 0;
w[15] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200s (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_s08 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = pws[gid].i[ 4];
w[ 5] = pws[gid].i[ 5];
w[ 6] = pws[gid].i[ 6];
w[ 7] = pws[gid].i[ 7];
w[ 8] = 0;
w[ 9] = 0;
w[10] = 0;
w[11] = 0;
w[12] = 0;
w[13] = 0;
w[14] = 0;
w[15] = 0;
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200s (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
__kernel void m00200_s16 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __constant u32x * words_buf_r, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
Initial commit
2015-12-04 14:47:52 +00:00
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w[16];
w[ 0] = pws[gid].i[ 0];
w[ 1] = pws[gid].i[ 1];
w[ 2] = pws[gid].i[ 2];
w[ 3] = pws[gid].i[ 3];
w[ 4] = pws[gid].i[ 4];
w[ 5] = pws[gid].i[ 5];
w[ 6] = pws[gid].i[ 6];
w[ 7] = pws[gid].i[ 7];
w[ 8] = pws[gid].i[ 8];
w[ 9] = pws[gid].i[ 9];
w[10] = pws[gid].i[10];
w[11] = pws[gid].i[11];
w[12] = pws[gid].i[12];
w[13] = pws[gid].i[13];
w[14] = pws[gid].i[14];
w[15] = pws[gid].i[15];
const u32 pw_len = pws[gid].pw_len;
/**
* main
*/
Fixed two major problems 1) SIMD code for all attack-mode Macro vector_accessible() was not refactored and missing completely. Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way. 2) SIMD code for attack-mode 0 With new SIMD code, apply_rules_vect() has to return u32 not u32x. This has massive impact on all *_a0 kernels. I've rewritten most of them. Deep testing using test.sh is still required. Some kernel need more fixes: - Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c - Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this. This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 16:18:54 +00:00
m00200s (w, pw_len, pws, rules_buf, combs_buf, words_buf_r, tmps, hooks, bitmaps_buf_s1_a, bitmaps_buf_s1_b, bitmaps_buf_s1_c, bitmaps_buf_s1_d, bitmaps_buf_s2_a, bitmaps_buf_s2_b, bitmaps_buf_s2_c, bitmaps_buf_s2_d, plains_buf, digests_buf, hashes_shown, salt_bufs, esalt_bufs, d_return_buf, d_scryptV_buf, bitmap_mask, bitmap_shift1, bitmap_shift2, salt_pos, loop_pos, loop_cnt, il_cnt, digests_cnt, digests_offset);
Initial commit
2015-12-04 14:47:52 +00:00
}
Reference in New Issue Copy Permalink