github/hashcat
1
Fork 0
mirror of https://github.com/hashcat/hashcat synced 2024-11-24 14:27:14 +01:00
Code Releases Activity

Workaround -m 7900 performance loss on AMD caused by CL1.2

Browse Source
This commit is contained in:
jsteube 2016-11-30 18:55:02 +01:00
parent d8cccfce48
commit 1eab457963
1 changed files with 20 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
OpenCL/m07900.cl
View File

@ -240,28 +240,22 @@ __kernel void m07900_loop (__global pw_t *pws, __global const kernel_rule_t *rul
if (gid >= gid_max) return;
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
w0[0] = pws[gid].i[ 0];
w0[1] = pws[gid].i[ 1];
w0[2] = pws[gid].i[ 2];
w0[3] = pws[gid].i[ 3];
u32 w1[4];
w1[0] = pws[gid].i[ 4];
w1[1] = pws[gid].i[ 5];
w1[2] = pws[gid].i[ 6];
w1[3] = pws[gid].i[ 7];
u32 w2[4];
w2[0] = pws[gid].i[ 8];
w2[1] = pws[gid].i[ 9];
w2[2] = pws[gid].i[10];
w2[3] = pws[gid].i[11];
u32 w3[4];
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
@ -292,24 +286,14 @@ __kernel void m07900_loop (__global pw_t *pws, __global const kernel_rule_t *rul
u32 block_len = (64 + pw_len);
u64 w[16];
u64 w_t[6];
w[ 0] = 0;
w[ 1] = 0;
w[ 2] = 0;
w[ 3] = 0;
w[ 4] = 0;
w[ 5] = 0;
w[ 6] = 0;
w[ 7] = 0;
w[ 8] = ((u64) swap32 (w0[0])) << 32 | (u64) swap32 (w0[1]);
w[ 9] = ((u64) swap32 (w0[2])) << 32 | (u64) swap32 (w0[3]);
w[10] = ((u64) swap32 (w1[0])) << 32 | (u64) swap32 (w1[1]);
w[11] = ((u64) swap32 (w1[2])) << 32 | (u64) swap32 (w1[3]);
w[12] = ((u64) swap32 (w2[0])) << 32 | (u64) swap32 (w2[1]);
w[13] = ((u64) swap32 (w2[2])) << 32 | (u64) swap32 (w2[3]);
w[14] = 0;
w[15] = block_len * 8;
w_t[0] = ((u64) swap32 (w0[0])) << 32 | (u64) swap32 (w0[1]);
w_t[1] = ((u64) swap32 (w0[2])) << 32 | (u64) swap32 (w0[3]);
w_t[2] = ((u64) swap32 (w1[0])) << 32 | (u64) swap32 (w1[1]);
w_t[3] = ((u64) swap32 (w1[2])) << 32 | (u64) swap32 (w1[3]);
w_t[4] = ((u64) swap32 (w2[0])) << 32 | (u64) swap32 (w2[1]);
w_t[5] = ((u64) swap32 (w2[2])) << 32 | (u64) swap32 (w2[3]);
/**
* init
@ -317,6 +301,8 @@ __kernel void m07900_loop (__global pw_t *pws, __global const kernel_rule_t *rul
for (u32 i = 0; i < loop_cnt; i++)
{
u64 w[16];
w[ 0] = digest[0];
w[ 1] = digest[1];
w[ 2] = digest[2];
@ -325,6 +311,14 @@ __kernel void m07900_loop (__global pw_t *pws, __global const kernel_rule_t *rul
w[ 5] = digest[5];
w[ 6] = digest[6];
w[ 7] = digest[7];
w[ 8] = w_t[0];
w[ 9] = w_t[1];
w[10] = w_t[2];
w[11] = w_t[3];
w[12] = w_t[4];
w[13] = w_t[5];
w[14] = 0;
w[15] = block_len * 8;
digest[0] = SHA512M_A;
digest[1] = SHA512M_B;