diff --git a/OpenCL/m00200_a0.cl b/OpenCL/m00200_a0.cl index 4d7d1ec66..34c690eef 100644 --- a/OpenCL/m00200_a0.cl +++ b/OpenCL/m00200_a0.cl @@ -40,18 +40,16 @@ __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, if (gid >= gid_max) return; u32 pw_buf0[4]; - - pw_buf0[0] = pws[gid].i[ 0]; - pw_buf0[1] = pws[gid].i[ 1]; - pw_buf0[2] = pws[gid].i[ 2]; - pw_buf0[3] = pws[gid].i[ 3]; - u32 pw_buf1[4]; - pw_buf1[0] = pws[gid].i[ 4]; - pw_buf1[1] = pws[gid].i[ 5]; - pw_buf1[2] = pws[gid].i[ 6]; - pw_buf1[3] = pws[gid].i[ 7]; + pw_buf0[0] = pws[gid].i[0]; + pw_buf0[1] = pws[gid].i[1]; + pw_buf0[2] = pws[gid].i[2]; + pw_buf0[3] = pws[gid].i[3]; + pw_buf1[0] = pws[gid].i[4]; + pw_buf1[1] = pws[gid].i[5]; + pw_buf1[2] = pws[gid].i[6]; + pw_buf1[3] = pws[gid].i[7]; const u32 pw_len = pws[gid].pw_len; @@ -166,18 +164,16 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, if (gid >= gid_max) return; u32 pw_buf0[4]; - - pw_buf0[0] = pws[gid].i[ 0]; - pw_buf0[1] = pws[gid].i[ 1]; - pw_buf0[2] = pws[gid].i[ 2]; - pw_buf0[3] = pws[gid].i[ 3]; - u32 pw_buf1[4]; - pw_buf1[0] = pws[gid].i[ 4]; - pw_buf1[1] = pws[gid].i[ 5]; - pw_buf1[2] = pws[gid].i[ 6]; - pw_buf1[3] = pws[gid].i[ 7]; + pw_buf0[0] = pws[gid].i[0]; + pw_buf0[1] = pws[gid].i[1]; + pw_buf0[2] = pws[gid].i[2]; + pw_buf0[3] = pws[gid].i[3]; + pw_buf1[0] = pws[gid].i[4]; + pw_buf1[1] = pws[gid].i[5]; + pw_buf1[2] = pws[gid].i[6]; + pw_buf1[3] = pws[gid].i[7]; const u32 pw_len = pws[gid].pw_len; diff --git a/OpenCL/m00200_a1.cl b/OpenCL/m00200_a1.cl index bbc053eb1..856fb5604 100644 --- a/OpenCL/m00200_a1.cl +++ b/OpenCL/m00200_a1.cl @@ -5,6 +5,9 @@ #define _MYSQL323_ +//incompatible +//#define NEW_SIMD_CODE + #include "include/constants.h" #include "include/kernel_vendor.h" @@ -16,9 +19,7 @@ #include "include/kernel_functions.c" #include "OpenCL/types_ocl.c" #include "OpenCL/common.c" - -#define COMPARE_S "OpenCL/check_single_comp4.c" -#define COMPARE_M "OpenCL/check_multi_comp4.c" +#include "OpenCL/simd.c" __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __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) { @@ -36,107 +37,122 @@ __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, if (gid >= gid_max) return; - u32 wordl0[4]; + u32 pw_buf0[4]; + u32 pw_buf1[4]; - wordl0[0] = pws[gid].i[ 0]; - wordl0[1] = pws[gid].i[ 1]; - wordl0[2] = pws[gid].i[ 2]; - wordl0[3] = pws[gid].i[ 3]; - - u32 wordl1[4]; - - wordl1[0] = pws[gid].i[ 4]; - wordl1[1] = pws[gid].i[ 5]; - wordl1[2] = pws[gid].i[ 6]; - wordl1[3] = pws[gid].i[ 7]; - - u32 wordl2[4]; - - wordl2[0] = 0; - wordl2[1] = 0; - wordl2[2] = 0; - wordl2[3] = 0; - - u32 wordl3[4]; - - wordl3[0] = 0; - wordl3[1] = 0; - wordl3[2] = 0; - wordl3[3] = 0; + pw_buf0[0] = pws[gid].i[0]; + pw_buf0[1] = pws[gid].i[1]; + pw_buf0[2] = pws[gid].i[2]; + pw_buf0[3] = pws[gid].i[3]; + pw_buf1[0] = pws[gid].i[4]; + pw_buf1[1] = pws[gid].i[5]; + pw_buf1[2] = pws[gid].i[6]; + pw_buf1[3] = pws[gid].i[7]; const u32 pw_l_len = pws[gid].pw_len; - if (combs_mode == COMBINATOR_MODE_BASE_RIGHT) - { - switch_buffer_by_offset_le (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len); - } - /** * loop */ - for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { - const u32 pw_r_len = combs_buf[il_pos].pw_len; + const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos); - const u32 pw_len = pw_l_len + pw_r_len; + const u32x pw_len = pw_l_len + pw_r_len; - u32 wordr0[4]; + /** + * concat password candidate + */ - wordr0[0] = combs_buf[il_pos].i[0]; - wordr0[1] = combs_buf[il_pos].i[1]; - wordr0[2] = combs_buf[il_pos].i[2]; - wordr0[3] = combs_buf[il_pos].i[3]; + u32x wordl0[4] = { 0 }; + u32x wordl1[4] = { 0 }; + u32x wordl2[4] = { 0 }; + u32x wordl3[4] = { 0 }; - u32 wordr1[4]; + wordl0[0] = pw_buf0[0]; + wordl0[1] = pw_buf0[1]; + wordl0[2] = pw_buf0[2]; + wordl0[3] = pw_buf0[3]; + wordl1[0] = pw_buf1[0]; + wordl1[1] = pw_buf1[1]; + wordl1[2] = pw_buf1[2]; + wordl1[3] = pw_buf1[3]; - wordr1[0] = combs_buf[il_pos].i[4]; - wordr1[1] = combs_buf[il_pos].i[5]; - wordr1[2] = combs_buf[il_pos].i[6]; - wordr1[3] = combs_buf[il_pos].i[7]; + u32x wordr0[4] = { 0 }; + u32x wordr1[4] = { 0 }; + u32x wordr2[4] = { 0 }; + u32x wordr3[4] = { 0 }; - u32 wordr2[4]; - - wordr2[0] = 0; - wordr2[1] = 0; - wordr2[2] = 0; - wordr2[3] = 0; - - u32 wordr3[4]; - - wordr3[0] = 0; - wordr3[1] = 0; - wordr3[2] = 0; - wordr3[3] = 0; + wordr0[0] = ix_create_combt (combs_buf, il_pos, 0); + wordr0[1] = ix_create_combt (combs_buf, il_pos, 1); + wordr0[2] = ix_create_combt (combs_buf, il_pos, 2); + wordr0[3] = ix_create_combt (combs_buf, il_pos, 3); + wordr1[0] = ix_create_combt (combs_buf, il_pos, 4); + wordr1[1] = ix_create_combt (combs_buf, il_pos, 5); + wordr1[2] = ix_create_combt (combs_buf, il_pos, 6); + wordr1[3] = ix_create_combt (combs_buf, il_pos, 7); if (combs_mode == COMBINATOR_MODE_BASE_LEFT) { - switch_buffer_by_offset_le (wordr0, wordr1, wordr2, wordr3, pw_l_len); + switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len); + } + else + { + switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len); } - u32 w_t[16]; + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; - w_t[ 0] = wordl0[0] | wordr0[0]; - w_t[ 1] = wordl0[1] | wordr0[1]; - w_t[ 2] = wordl0[2] | wordr0[2]; - w_t[ 3] = wordl0[3] | wordr0[3]; - w_t[ 4] = wordl1[0] | wordr1[0]; - w_t[ 5] = wordl1[1] | wordr1[1]; - w_t[ 6] = wordl1[2] | wordr1[2]; - w_t[ 7] = wordl1[3] | wordr1[3]; - w_t[ 8] = wordl2[0] | wordr2[0]; - w_t[ 9] = wordl2[1] | wordr2[1]; - w_t[10] = wordl2[2] | wordr2[2]; - w_t[11] = wordl2[3] | wordr2[3]; - w_t[12] = wordl3[0] | wordr3[0]; - w_t[13] = wordl3[1] | wordr3[1]; - w_t[14] = wordl3[2] | wordr3[2]; + w0[0] = wordl0[0] | wordr0[0]; + w0[1] = wordl0[1] | wordr0[1]; + w0[2] = wordl0[2] | wordr0[2]; + w0[3] = wordl0[3] | wordr0[3]; + w1[0] = wordl1[0] | wordr1[0]; + w1[1] = wordl1[1] | wordr1[1]; + w1[2] = wordl1[2] | wordr1[2]; + w1[3] = wordl1[3] | wordr1[3]; + w2[0] = wordl2[0] | wordr2[0]; + w2[1] = wordl2[1] | wordr2[1]; + w2[2] = wordl2[2] | wordr2[2]; + w2[3] = wordl2[3] | wordr2[3]; + w3[0] = wordl3[0] | wordr3[0]; + w3[1] = wordl3[1] | wordr3[1]; + w3[2] = pw_len * 8; + w3[3] = 0; + + /** + * mysql + */ + + u32x w_t[16]; + + w_t[ 0] = w0[0]; + w_t[ 1] = w0[1]; + w_t[ 2] = w0[2]; + w_t[ 3] = w0[3]; + w_t[ 4] = w1[0]; + w_t[ 5] = w1[1]; + w_t[ 6] = w1[2]; + w_t[ 7] = w1[3]; + w_t[ 8] = 0; + w_t[ 9] = 0; + w_t[10] = 0; + w_t[11] = 0; + w_t[12] = 0; + w_t[13] = 0; + w_t[14] = 0; w_t[15] = 0; - u32 a = MYSQL323_A; - u32 b = MYSQL323_B; + u32x a = MYSQL323_A; + u32x b = MYSQL323_B; + u32x c = 0; + u32x d = 0; - u32 add = 7; + u32x add = 7; #define ROUND(v) \ { \ @@ -150,7 +166,7 @@ __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, for (i = 0, j = 0; i <= (int) pw_len - 4; i += 4, j += 1) { - const u32 wj = w_t[j]; + const u32x wj = w_t[j]; ROUND ((wj >> 0) & 0xff); ROUND ((wj >> 8) & 0xff); @@ -158,7 +174,7 @@ __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, ROUND ((wj >> 24) & 0xff); } - const u32 wj = w_t[j]; + const u32x wj = w_t[j]; const u32 left = pw_len - i; @@ -181,12 +197,7 @@ __kernel void m00200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, a &= 0x7fffffff; b &= 0x7fffffff; - const u32 r0 = a; - const u32 r1 = b; - const u32 r2 = 0; - const u32 r3 = 0; - - #include COMPARE_M + COMPARE_M_SIMD (a, b, c, d); } } @@ -214,41 +225,20 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, if (gid >= gid_max) return; - u32 wordl0[4]; + u32 pw_buf0[4]; + u32 pw_buf1[4]; - wordl0[0] = pws[gid].i[ 0]; - wordl0[1] = pws[gid].i[ 1]; - wordl0[2] = pws[gid].i[ 2]; - wordl0[3] = pws[gid].i[ 3]; - - u32 wordl1[4]; - - wordl1[0] = pws[gid].i[ 4]; - wordl1[1] = pws[gid].i[ 5]; - wordl1[2] = pws[gid].i[ 6]; - wordl1[3] = pws[gid].i[ 7]; - - u32 wordl2[4]; - - wordl2[0] = 0; - wordl2[1] = 0; - wordl2[2] = 0; - wordl2[3] = 0; - - u32 wordl3[4]; - - wordl3[0] = 0; - wordl3[1] = 0; - wordl3[2] = 0; - wordl3[3] = 0; + pw_buf0[0] = pws[gid].i[0]; + pw_buf0[1] = pws[gid].i[1]; + pw_buf0[2] = pws[gid].i[2]; + pw_buf0[3] = pws[gid].i[3]; + pw_buf1[0] = pws[gid].i[4]; + pw_buf1[1] = pws[gid].i[5]; + pw_buf1[2] = pws[gid].i[6]; + pw_buf1[3] = pws[gid].i[7]; const u32 pw_l_len = pws[gid].pw_len; - if (combs_mode == COMBINATOR_MODE_BASE_RIGHT) - { - switch_buffer_by_offset_le (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len); - } - /** * digest */ @@ -265,68 +255,104 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, * loop */ - for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { - const u32 pw_r_len = combs_buf[il_pos].pw_len; + const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos); - const u32 pw_len = pw_l_len + pw_r_len; + const u32x pw_len = pw_l_len + pw_r_len; - u32 wordr0[4]; + /** + * concat password candidate + */ - wordr0[0] = combs_buf[il_pos].i[0]; - wordr0[1] = combs_buf[il_pos].i[1]; - wordr0[2] = combs_buf[il_pos].i[2]; - wordr0[3] = combs_buf[il_pos].i[3]; + u32x wordl0[4] = { 0 }; + u32x wordl1[4] = { 0 }; + u32x wordl2[4] = { 0 }; + u32x wordl3[4] = { 0 }; - u32 wordr1[4]; + wordl0[0] = pw_buf0[0]; + wordl0[1] = pw_buf0[1]; + wordl0[2] = pw_buf0[2]; + wordl0[3] = pw_buf0[3]; + wordl1[0] = pw_buf1[0]; + wordl1[1] = pw_buf1[1]; + wordl1[2] = pw_buf1[2]; + wordl1[3] = pw_buf1[3]; - wordr1[0] = combs_buf[il_pos].i[4]; - wordr1[1] = combs_buf[il_pos].i[5]; - wordr1[2] = combs_buf[il_pos].i[6]; - wordr1[3] = combs_buf[il_pos].i[7]; + u32x wordr0[4] = { 0 }; + u32x wordr1[4] = { 0 }; + u32x wordr2[4] = { 0 }; + u32x wordr3[4] = { 0 }; - u32 wordr2[4]; - - wordr2[0] = 0; - wordr2[1] = 0; - wordr2[2] = 0; - wordr2[3] = 0; - - u32 wordr3[4]; - - wordr3[0] = 0; - wordr3[1] = 0; - wordr3[2] = 0; - wordr3[3] = 0; + wordr0[0] = ix_create_combt (combs_buf, il_pos, 0); + wordr0[1] = ix_create_combt (combs_buf, il_pos, 1); + wordr0[2] = ix_create_combt (combs_buf, il_pos, 2); + wordr0[3] = ix_create_combt (combs_buf, il_pos, 3); + wordr1[0] = ix_create_combt (combs_buf, il_pos, 4); + wordr1[1] = ix_create_combt (combs_buf, il_pos, 5); + wordr1[2] = ix_create_combt (combs_buf, il_pos, 6); + wordr1[3] = ix_create_combt (combs_buf, il_pos, 7); if (combs_mode == COMBINATOR_MODE_BASE_LEFT) { - switch_buffer_by_offset_le (wordr0, wordr1, wordr2, wordr3, pw_l_len); + switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len); + } + else + { + switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len); } - u32 w_t[16]; + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; - w_t[ 0] = wordl0[0] | wordr0[0]; - w_t[ 1] = wordl0[1] | wordr0[1]; - w_t[ 2] = wordl0[2] | wordr0[2]; - w_t[ 3] = wordl0[3] | wordr0[3]; - w_t[ 4] = wordl1[0] | wordr1[0]; - w_t[ 5] = wordl1[1] | wordr1[1]; - w_t[ 6] = wordl1[2] | wordr1[2]; - w_t[ 7] = wordl1[3] | wordr1[3]; - w_t[ 8] = wordl2[0] | wordr2[0]; - w_t[ 9] = wordl2[1] | wordr2[1]; - w_t[10] = wordl2[2] | wordr2[2]; - w_t[11] = wordl2[3] | wordr2[3]; - w_t[12] = wordl3[0] | wordr3[0]; - w_t[13] = wordl3[1] | wordr3[1]; - w_t[14] = wordl3[2] | wordr3[2]; + w0[0] = wordl0[0] | wordr0[0]; + w0[1] = wordl0[1] | wordr0[1]; + w0[2] = wordl0[2] | wordr0[2]; + w0[3] = wordl0[3] | wordr0[3]; + w1[0] = wordl1[0] | wordr1[0]; + w1[1] = wordl1[1] | wordr1[1]; + w1[2] = wordl1[2] | wordr1[2]; + w1[3] = wordl1[3] | wordr1[3]; + w2[0] = wordl2[0] | wordr2[0]; + w2[1] = wordl2[1] | wordr2[1]; + w2[2] = wordl2[2] | wordr2[2]; + w2[3] = wordl2[3] | wordr2[3]; + w3[0] = wordl3[0] | wordr3[0]; + w3[1] = wordl3[1] | wordr3[1]; + w3[2] = pw_len * 8; + w3[3] = 0; + + /** + * mysql + */ + + u32x w_t[16]; + + w_t[ 0] = w0[0]; + w_t[ 1] = w0[1]; + w_t[ 2] = w0[2]; + w_t[ 3] = w0[3]; + w_t[ 4] = w1[0]; + w_t[ 5] = w1[1]; + w_t[ 6] = w1[2]; + w_t[ 7] = w1[3]; + w_t[ 8] = 0; + w_t[ 9] = 0; + w_t[10] = 0; + w_t[11] = 0; + w_t[12] = 0; + w_t[13] = 0; + w_t[14] = 0; w_t[15] = 0; - u32 a = MYSQL323_A; - u32 b = MYSQL323_B; + u32x a = MYSQL323_A; + u32x b = MYSQL323_B; + u32x c = 0; + u32x d = 0; - u32 add = 7; + u32x add = 7; #define ROUND(v) \ { \ @@ -340,7 +366,7 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, for (i = 0, j = 0; i <= (int) pw_len - 4; i += 4, j += 1) { - const u32 wj = w_t[j]; + const u32x wj = w_t[j]; ROUND ((wj >> 0) & 0xff); ROUND ((wj >> 8) & 0xff); @@ -348,7 +374,7 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, ROUND ((wj >> 24) & 0xff); } - const u32 wj = w_t[j]; + const u32x wj = w_t[j]; const u32 left = pw_len - i; @@ -371,12 +397,7 @@ __kernel void m00200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, a &= 0x7fffffff; b &= 0x7fffffff; - const u32 r0 = a; - const u32 r1 = b; - const u32 r2 = 0; - const u32 r3 = 0; - - #include COMPARE_S + COMPARE_S_SIMD (a, b, c, d); } } diff --git a/OpenCL/m00200_a3.cl b/OpenCL/m00200_a3.cl index f0fae5479..bdcadac2c 100644 --- a/OpenCL/m00200_a3.cl +++ b/OpenCL/m00200_a3.cl @@ -41,6 +41,25 @@ static void m00200m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k const u32x w0 = w0l | w0r; + u32x w_t[16]; + + w_t[ 0] = w0; + w_t[ 1] = w[ 1]; + w_t[ 2] = w[ 2]; + w_t[ 3] = w[ 3]; + w_t[ 4] = w[ 4]; + w_t[ 5] = w[ 5]; + w_t[ 6] = w[ 6]; + w_t[ 7] = w[ 7]; + w_t[ 8] = w[ 8]; + w_t[ 9] = w[ 9]; + w_t[10] = w[10]; + w_t[11] = w[11]; + w_t[12] = w[12]; + w_t[13] = w[13]; + w_t[14] = w[14]; + w_t[15] = w[15]; + u32x a = MYSQL323_A; u32x b = MYSQL323_B; u32x c = 0; @@ -55,35 +74,12 @@ static void m00200m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k add += v; \ } - if (pw_len >= 4) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - ROUND ((w0 >> 16) & 0xff); - ROUND ((w0 >> 24) & 0xff); - } - else if (pw_len == 3) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - ROUND ((w0 >> 16) & 0xff); - } - else if (pw_len == 2) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - } - else if (pw_len == 1) - { - ROUND ((w0 >> 0) & 0xff); - } - int i; int j; - for (i = 4, j = 1; i <= (int) pw_len - 4; i += 4, j += 1) + for (i = 0, j = 0; i <= (int) pw_len - 4; i += 4, j += 1) { - const u32x wj = w[j]; + const u32x wj = w_t[j]; ROUND ((wj >> 0) & 0xff); ROUND ((wj >> 8) & 0xff); @@ -91,7 +87,7 @@ static void m00200m (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k ROUND ((wj >> 24) & 0xff); } - const u32x wj = w[j]; + const u32x wj = w_t[j]; const u32 left = pw_len - i; @@ -151,6 +147,25 @@ static void m00200s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k const u32x w0 = w0l | w0r; + u32x w_t[16]; + + w_t[ 0] = w0; + w_t[ 1] = w[ 1]; + w_t[ 2] = w[ 2]; + w_t[ 3] = w[ 3]; + w_t[ 4] = w[ 4]; + w_t[ 5] = w[ 5]; + w_t[ 6] = w[ 6]; + w_t[ 7] = w[ 7]; + w_t[ 8] = w[ 8]; + w_t[ 9] = w[ 9]; + w_t[10] = w[10]; + w_t[11] = w[11]; + w_t[12] = w[12]; + w_t[13] = w[13]; + w_t[14] = w[14]; + w_t[15] = w[15]; + u32x a = MYSQL323_A; u32x b = MYSQL323_B; u32x c = 0; @@ -165,35 +180,12 @@ static void m00200s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k add += v; \ } - if (pw_len >= 4) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - ROUND ((w0 >> 16) & 0xff); - ROUND ((w0 >> 24) & 0xff); - } - else if (pw_len == 3) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - ROUND ((w0 >> 16) & 0xff); - } - else if (pw_len == 2) - { - ROUND ((w0 >> 0) & 0xff); - ROUND ((w0 >> 8) & 0xff); - } - else if (pw_len == 1) - { - ROUND ((w0 >> 0) & 0xff); - } - int i; int j; - for (i = 4, j = 1; i <= (int) pw_len - 4; i += 4, j += 1) + for (i = 0, j = 0; i <= (int) pw_len - 4; i += 4, j += 1) { - const u32x wj = w[j]; + const u32x wj = w_t[j]; ROUND ((wj >> 0) & 0xff); ROUND ((wj >> 8) & 0xff); @@ -201,7 +193,7 @@ static void m00200s (u32 w[16], const u32 pw_len, __global pw_t *pws, __global k ROUND ((wj >> 24) & 0xff); } - const u32x wj = w[j]; + const u32x wj = w_t[j]; const u32 left = pw_len - i;