arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2024-11-23 00:28:11 +00:00
Code Issues Releases Wiki Activity

Converted to new SIMD: -m 2810 -a 0

Browse Source
This commit is contained in:
Jens Steube 2016-02-05 21:23:39 +01:00
parent ef3afbab9d
commit 605f8f07fb
1 changed files with 81 additions and 140 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

221
OpenCL/m02810_a0.cl
View File

@ -5,6 +5,8 @@
#define _MD5_
#define NEW_SIMD_CODE
#include "include/constants.h"
#include "include/kernel_vendor.h"
@ -18,9 +20,7 @@
#include "OpenCL/common.c"
#include "include/rp_kernel.h"
#include "OpenCL/rp.c"
#define COMPARE_S "OpenCL/check_single_comp4.c"
#define COMPARE_M "OpenCL/check_multi_comp4.c"
#include "OpenCL/simd.c"
#if VECT_SIZE == 1
#define uint_to_hex_lower8(i) (u32x) (l_bin2asc[(i)])
@ -107,46 +107,23 @@ __kernel void m02810_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
* loop
*/
for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
for (u32 il_pos = 0; il_pos < rules_cnt; il_pos += VECT_SIZE)
{
u32 w0[4];
u32x w0[4] = { 0 };
u32x w1[4] = { 0 };
u32x w2[4] = { 0 };
u32x w3[4] = { 0 };
w0[0] = pw_buf0[0];
w0[1] = pw_buf0[1];
w0[2] = pw_buf0[2];
w0[3] = pw_buf0[3];
u32 w1[4];
w1[0] = pw_buf1[0];
w1[1] = pw_buf1[1];
w1[2] = pw_buf1[2];
w1[3] = pw_buf1[3];
u32 w2[4];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
u32 w3[4];
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
const u32 out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
append_0x80_2x4 (w0, w1, out_len);
w3[2] = out_len * 8;
u32 a = MD5M_A;
u32 b = MD5M_B;
u32 c = MD5M_C;
u32 d = MD5M_D;
u32x a = MD5M_A;
u32x b = MD5M_B;
u32x c = MD5M_C;
u32x d = MD5M_D;
MD5_STEP (MD5_Fo, a, b, c, d, w0[0], MD5C00, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, w0[1], MD5C01, MD5S01);
@ -221,31 +198,31 @@ __kernel void m02810_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
c += MD5M_C;
d += MD5M_D;
const u32 w0_t = s[0];
const u32 w1_t = s[1];
const u32 w2_t = s[2];
const u32 w3_t = s[3];
const u32 w4_t = s[4];
const u32 w5_t = s[5];
const u32 w6_t = s[6];
const u32 w7_t = s[7];
const u32x w0_t = s[0];
const u32x w1_t = s[1];
const u32x w2_t = s[2];
const u32x w3_t = s[3];
const u32x w4_t = s[4];
const u32x w5_t = s[5];
const u32x w6_t = s[6];
const u32x w7_t = s[7];
const u32 w8_t = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
const u32 w9_t = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
const u32 wa_t = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
const u32 wb_t = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
const u32 wc_t = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
const u32 wd_t = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
const u32 we_t = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
const u32 wf_t = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
const u32x w8_t = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
const u32x w9_t = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
const u32x wa_t = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
const u32x wb_t = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
const u32x wc_t = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
const u32x wd_t = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
const u32x we_t = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
const u32x wf_t = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
a = MD5M_A;
b = MD5M_B;
@ -320,10 +297,10 @@ __kernel void m02810_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
MD5_STEP (MD5_I , c, d, a, b, w2_t, MD5C3e, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, w9_t, MD5C3f, MD5S33);
const u32 r_a = a + MD5M_A;
const u32 r_b = b + MD5M_B;
const u32 r_c = c + MD5M_C;
const u32 r_d = d + MD5M_D;
const u32x r_a = a + MD5M_A;
const u32x r_b = b + MD5M_B;
const u32x r_c = c + MD5M_C;
const u32x r_d = d + MD5M_D;
a = r_a;
b = r_b;
@ -403,12 +380,7 @@ __kernel void m02810_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
c += r_c;
d += r_d;
const u32 r0 = a;
const u32 r1 = d;
const u32 r2 = c;
const u32 r3 = b;
#include COMPARE_M
COMPARE_M_SIMD (a, d, c, b);
}
}
@ -503,46 +475,23 @@ __kernel void m02810_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
* loop
*/
for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
for (u32 il_pos = 0; il_pos < rules_cnt; il_pos += VECT_SIZE)
{
u32 w0[4];
u32x w0[4] = { 0 };
u32x w1[4] = { 0 };
u32x w2[4] = { 0 };
u32x w3[4] = { 0 };
w0[0] = pw_buf0[0];
w0[1] = pw_buf0[1];
w0[2] = pw_buf0[2];
w0[3] = pw_buf0[3];
u32 w1[4];
w1[0] = pw_buf1[0];
w1[1] = pw_buf1[1];
w1[2] = pw_buf1[2];
w1[3] = pw_buf1[3];
u32 w2[4];
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
u32 w3[4];
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
const u32 out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
append_0x80_2x4 (w0, w1, out_len);
w3[2] = out_len * 8;
u32 a = MD5M_A;
u32 b = MD5M_B;
u32 c = MD5M_C;
u32 d = MD5M_D;
u32x a = MD5M_A;
u32x b = MD5M_B;
u32x c = MD5M_C;
u32x d = MD5M_D;
MD5_STEP (MD5_Fo, a, b, c, d, w0[0], MD5C00, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, w0[1], MD5C01, MD5S01);
@ -617,31 +566,31 @@ __kernel void m02810_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
c += MD5M_C;
d += MD5M_D;
const u32 w0_t = s[0];
const u32 w1_t = s[1];
const u32 w2_t = s[2];
const u32 w3_t = s[3];
const u32 w4_t = s[4];
const u32 w5_t = s[5];
const u32 w6_t = s[6];
const u32 w7_t = s[7];
const u32x w0_t = s[0];
const u32x w1_t = s[1];
const u32x w2_t = s[2];
const u32x w3_t = s[3];
const u32x w4_t = s[4];
const u32x w5_t = s[5];
const u32x w6_t = s[6];
const u32x w7_t = s[7];
const u32 w8_t = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
const u32 w9_t = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
const u32 wa_t = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
const u32 wb_t = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
const u32 wc_t = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
const u32 wd_t = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
const u32 we_t = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
const u32 wf_t = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
const u32x w8_t = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
const u32x w9_t = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
const u32x wa_t = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
const u32x wb_t = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
const u32x wc_t = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
const u32x wd_t = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
const u32x we_t = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
const u32x wf_t = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
a = MD5M_A;
b = MD5M_B;
@ -716,10 +665,10 @@ __kernel void m02810_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
MD5_STEP (MD5_I , c, d, a, b, w2_t, MD5C3e, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, w9_t, MD5C3f, MD5S33);
const u32 r_a = a + MD5M_A;
const u32 r_b = b + MD5M_B;
const u32 r_c = c + MD5M_C;
const u32 r_d = d + MD5M_D;
const u32x r_a = a + MD5M_A;
const u32x r_b = b + MD5M_B;
const u32x r_c = c + MD5M_C;
const u32x r_d = d + MD5M_D;
a = r_a;
b = r_b;
@ -790,9 +739,6 @@ __kernel void m02810_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
MD5_STEP0(MD5_I , c, d, a, b, MD5C3a, MD5S32);
MD5_STEP0(MD5_I , b, c, d, a, MD5C3b, MD5S33);
MD5_STEP0(MD5_I , a, b, c, d, MD5C3c, MD5S30);
if (allx ((a + r_a) != search[0])) continue;
MD5_STEP0(MD5_I , d, a, b, c, MD5C3d, MD5S31);
MD5_STEP0(MD5_I , c, d, a, b, MD5C3e, MD5S32);
MD5_STEP0(MD5_I , b, c, d, a, MD5C3f, MD5S33);
@ -802,12 +748,7 @@ __kernel void m02810_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf,
c += r_c;
d += r_d;
const u32 r0 = a;
const u32 r1 = d;
const u32 r2 = c;
const u32 r3 = b;
#include COMPARE_S
COMPARE_S_SIMD (a, d, c, b);
}
}