From 2348fb3f3603d5715c3e75678b403591ddd5d049 Mon Sep 17 00:00:00 2001 From: Jens Steube Date: Tue, 12 Apr 2016 08:53:08 +0200 Subject: [PATCH] Cleanup -m 51xx kernels to latest standard --- OpenCL/m05100_a0.cl | 58 +++++--- OpenCL/m05100_a1.cl | 350 ++++++++++++++++++-------------------------- OpenCL/m05100_a3.cl | 336 +++++++++++++++++++++++------------------- 3 files changed, 366 insertions(+), 378 deletions(-) diff --git a/OpenCL/m05100_a0.cl b/OpenCL/m05100_a0.cl index 15488680c..9f991831a 100644 --- a/OpenCL/m05100_a0.cl +++ b/OpenCL/m05100_a0.cl @@ -39,21 +39,31 @@ __kernel void m05100_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; + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[DGST_R0], + digests_buf[digests_offset].digest_buf[DGST_R1], + digests_buf[digests_offset].digest_buf[DGST_R2], + digests_buf[digests_offset].digest_buf[DGST_R3] + }; + /** * loop */ @@ -70,6 +80,11 @@ __kernel void m05100_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, append_0x80_2x4_VV (w0, w1, out_len); w3[2] = out_len * 8; + w3[3] = 0; + + /** + * md5 + */ u32x a = MD5M_A; u32x b = MD5M_B; @@ -185,18 +200,16 @@ __kernel void m05100_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; @@ -228,6 +241,11 @@ __kernel void m05100_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, append_0x80_2x4_VV (w0, w1, out_len); w3[2] = out_len * 8; + w3[3] = 0; + + /** + * md5 + */ u32x a = MD5M_A; u32x b = MD5M_B; diff --git a/OpenCL/m05100_a1.cl b/OpenCL/m05100_a1.cl index 1798d2c79..d9eb91d6a 100644 --- a/OpenCL/m05100_a1.cl +++ b/OpenCL/m05100_a1.cl @@ -5,6 +5,8 @@ #define _MD5H_ +#define NEW_SIMD_CODE + #include "include/constants.h" #include "include/kernel_vendor.h" @@ -16,9 +18,7 @@ #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 m05100_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) { @@ -36,118 +36,101 @@ __kernel void m05100_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, if (gid >= gid_max) return; - u32 wordl0[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 pw_buf0[4]; + u32 pw_buf1[4]; - 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) - { - append_0x80_2x4 (wordl0, wordl1, pw_l_len); - - 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 u32 pw_len = pw_l_len + pw_r_len; - - u32 wordr0[4]; - - 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]; - - u32 wordr1[4]; - - 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]; - - 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; + const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos); + + const u32x pw_len = pw_l_len + pw_r_len; + + /** + * concat password candidate + */ + + u32x wordl0[4] = { 0 }; + u32x wordl1[4] = { 0 }; + u32x wordl2[4] = { 0 }; + u32x wordl3[4] = { 0 }; + + 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]; + + u32x wordr0[4] = { 0 }; + u32x wordr1[4] = { 0 }; + u32x wordr2[4] = { 0 }; + u32x wordr3[4] = { 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 w0[4]; + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; w0[0] = wordl0[0] | wordr0[0]; w0[1] = wordl0[1] | wordr0[1]; w0[2] = wordl0[2] | wordr0[2]; w0[3] = wordl0[3] | wordr0[3]; - - u32 w1[4]; - w1[0] = wordl1[0] | wordr1[0]; w1[1] = wordl1[1] | wordr1[1]; w1[2] = wordl1[2] | wordr1[2]; w1[3] = wordl1[3] | wordr1[3]; - - u32 w2[4]; - w2[0] = wordl2[0] | wordr2[0]; w2[1] = wordl2[1] | wordr2[1]; w2[2] = wordl2[2] | wordr2[2]; w2[3] = wordl2[3] | wordr2[3]; - - u32 w3[4]; - w3[0] = wordl3[0] | wordr3[0]; w3[1] = wordl3[1] | wordr3[1]; w3[2] = pw_len * 8; w3[3] = 0; - u32 a = MD5M_A; - u32 b = MD5M_B; - u32 c = MD5M_C; - u32 d = MD5M_D; + /** + * md5 + */ + + 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); @@ -222,32 +205,14 @@ __kernel void m05100_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, c += MD5M_C; d += MD5M_D; - { - const u32 r0 = a; - const u32 r1 = b; - const u32 r2 = 0; - const u32 r3 = 0; + u32x e = 0; + u32x f = 0; - #include COMPARE_M - } + COMPARE_M_SIMD (a, b, e, f); - { - const u32 r0 = b; - const u32 r1 = c; - const u32 r2 = 0; - const u32 r3 = 0; + COMPARE_M_SIMD (b, c, e, f); - #include COMPARE_M - } - - { - const u32 r0 = c; - const u32 r1 = d; - const u32 r2 = 0; - const u32 r3 = 0; - - #include COMPARE_M - } + COMPARE_M_SIMD (c, d, e, f); } } @@ -275,33 +240,19 @@ __kernel void m05100_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]; + 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]; - 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; + const u32 pw_l_len = pws[gid].pw_len; /** * digest @@ -315,90 +266,87 @@ __kernel void m05100_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, digests_buf[digests_offset].digest_buf[DGST_R3] }; - const u32 pw_l_len = pws[gid].pw_len; - - if (combs_mode == COMBINATOR_MODE_BASE_RIGHT) - { - append_0x80_2x4 (wordl0, wordl1, pw_l_len); - - 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 u32 pw_len = pw_l_len + pw_r_len; - - u32 wordr0[4]; - - 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]; - - u32 wordr1[4]; - - 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]; - - 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; + const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos); + + const u32x pw_len = pw_l_len + pw_r_len; + + /** + * concat password candidate + */ + + u32x wordl0[4] = { 0 }; + u32x wordl1[4] = { 0 }; + u32x wordl2[4] = { 0 }; + u32x wordl3[4] = { 0 }; + + 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]; + + u32x wordr0[4] = { 0 }; + u32x wordr1[4] = { 0 }; + u32x wordr2[4] = { 0 }; + u32x wordr3[4] = { 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 w0[4]; + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; w0[0] = wordl0[0] | wordr0[0]; w0[1] = wordl0[1] | wordr0[1]; w0[2] = wordl0[2] | wordr0[2]; w0[3] = wordl0[3] | wordr0[3]; - - u32 w1[4]; - w1[0] = wordl1[0] | wordr1[0]; w1[1] = wordl1[1] | wordr1[1]; w1[2] = wordl1[2] | wordr1[2]; w1[3] = wordl1[3] | wordr1[3]; - - u32 w2[4]; - w2[0] = wordl2[0] | wordr2[0]; w2[1] = wordl2[1] | wordr2[1]; w2[2] = wordl2[2] | wordr2[2]; w2[3] = wordl2[3] | wordr2[3]; - - u32 w3[4]; - w3[0] = wordl3[0] | wordr3[0]; w3[1] = wordl3[1] | wordr3[1]; w3[2] = pw_len * 8; w3[3] = 0; - u32 a = MD5M_A; - u32 b = MD5M_B; - u32 c = MD5M_C; - u32 d = MD5M_D; + /** + * md5 + */ + + 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); @@ -473,32 +421,14 @@ __kernel void m05100_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, c += MD5M_C; d += MD5M_D; - { - const u32 r0 = a; - const u32 r1 = b; - const u32 r2 = 0; - const u32 r3 = 0; + u32x e = 0; + u32x f = 0; - #include COMPARE_S - } + COMPARE_S_SIMD (a, b, e, f); - { - const u32 r0 = b; - const u32 r1 = c; - const u32 r2 = 0; - const u32 r3 = 0; + COMPARE_S_SIMD (b, c, e, f); - #include COMPARE_S - } - - { - const u32 r0 = c; - const u32 r1 = d; - const u32 r2 = 0; - const u32 r3 = 0; - - #include COMPARE_S - } + COMPARE_S_SIMD (c, d, e, f); } } diff --git a/OpenCL/m05100_a3.cl b/OpenCL/m05100_a3.cl index a23d8678c..956157fe1 100644 --- a/OpenCL/m05100_a3.cl +++ b/OpenCL/m05100_a3.cl @@ -29,17 +29,11 @@ static void m05100m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le const u32 gid = get_global_id (0); const u32 lid = get_local_id (0); - /** - * base - */ - - w3[2] = pw_len * 8; - /** * loop */ - u32 w0l = w0[0]; + const u32 w0l = w0[0]; for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { @@ -47,78 +41,104 @@ static void m05100m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le const u32x w0lr = w0l | w0r; + u32x t0[4]; + u32x t1[4]; + u32x t2[4]; + u32x t3[4]; + + t0[0] = w0lr; + t0[1] = w0[1]; + t0[2] = w0[2]; + t0[3] = w0[3]; + t1[0] = w1[0]; + t1[1] = w1[1]; + t1[2] = w1[2]; + t1[3] = w1[3]; + t2[0] = w2[0]; + t2[1] = w2[1]; + t2[2] = w2[2]; + t2[3] = w2[3]; + t3[0] = w3[0]; + t3[1] = w3[1]; + t3[2] = pw_len * 8; + t3[3] = 0; + + /** + * md5 + */ + u32x a = MD5M_A; u32x b = MD5M_B; u32x c = MD5M_C; u32x d = MD5M_D; - MD5_STEP (MD5_Fo, a, b, c, d, w0lr, MD5C00, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w0[1], MD5C01, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w0[2], MD5C02, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w0[3], MD5C03, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w1[0], MD5C04, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w1[1], MD5C05, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w1[2], MD5C06, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w1[3], MD5C07, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w2[0], MD5C08, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w2[1], MD5C09, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w2[2], MD5C0a, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w2[3], MD5C0b, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w3[0], MD5C0c, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w3[1], MD5C0d, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w3[2], MD5C0e, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w3[3], MD5C0f, MD5S03); - - MD5_STEP (MD5_Go, a, b, c, d, w0[1], MD5C10, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w1[2], MD5C11, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w2[3], MD5C12, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w0lr, MD5C13, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w1[1], MD5C14, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w2[2], MD5C15, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w3[3], MD5C16, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w1[0], MD5C17, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w2[1], MD5C18, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w3[2], MD5C19, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w0[3], MD5C1a, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w2[0], MD5C1b, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w3[1], MD5C1c, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w0[2], MD5C1d, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w1[3], MD5C1e, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w3[0], MD5C1f, MD5S13); - - MD5_STEP (MD5_H , a, b, c, d, w1[1], MD5C20, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w2[0], MD5C21, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w2[3], MD5C22, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w3[2], MD5C23, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w0[1], MD5C24, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w1[0], MD5C25, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w1[3], MD5C26, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w2[2], MD5C27, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w3[1], MD5C28, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w0lr, MD5C29, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w0[3], MD5C2a, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w1[2], MD5C2b, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w2[1], MD5C2c, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w3[0], MD5C2d, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w3[3], MD5C2e, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w0[2], MD5C2f, MD5S23); - - MD5_STEP (MD5_I , a, b, c, d, w0lr, MD5C30, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w1[3], MD5C31, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w3[2], MD5C32, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w1[1], MD5C33, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w3[0], MD5C34, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w0[3], MD5C35, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w2[2], MD5C36, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w0[1], MD5C37, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w2[0], MD5C38, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w3[3], MD5C39, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w1[2], MD5C3a, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w3[1], MD5C3b, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w1[0], MD5C3c, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w2[3], MD5C3d, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w0[2], MD5C3e, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w2[1], MD5C3f, MD5S33); + MD5_STEP (MD5_Fo, a, b, c, d, t0[0], MD5C00, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t0[1], MD5C01, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t0[2], MD5C02, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t0[3], MD5C03, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t1[0], MD5C04, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t1[1], MD5C05, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t1[2], MD5C06, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t1[3], MD5C07, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t2[0], MD5C08, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t2[1], MD5C09, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t2[2], MD5C0a, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t2[3], MD5C0b, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t3[0], MD5C0c, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t3[1], MD5C0d, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t3[2], MD5C0e, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t3[3], MD5C0f, MD5S03); + + MD5_STEP (MD5_Go, a, b, c, d, t0[1], MD5C10, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t1[2], MD5C11, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t2[3], MD5C12, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t0[0], MD5C13, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t1[1], MD5C14, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t2[2], MD5C15, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t3[3], MD5C16, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t1[0], MD5C17, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t2[1], MD5C18, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t3[2], MD5C19, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t0[3], MD5C1a, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t2[0], MD5C1b, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t3[1], MD5C1c, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t0[2], MD5C1d, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t1[3], MD5C1e, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t3[0], MD5C1f, MD5S13); + + MD5_STEP (MD5_H , a, b, c, d, t1[1], MD5C20, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t2[0], MD5C21, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t2[3], MD5C22, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t3[2], MD5C23, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t0[1], MD5C24, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t1[0], MD5C25, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t1[3], MD5C26, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t2[2], MD5C27, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t3[1], MD5C28, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t0[0], MD5C29, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t0[3], MD5C2a, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t1[2], MD5C2b, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t2[1], MD5C2c, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t3[0], MD5C2d, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t3[3], MD5C2e, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t0[2], MD5C2f, MD5S23); + + MD5_STEP (MD5_I , a, b, c, d, t0[0], MD5C30, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t1[3], MD5C31, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t3[2], MD5C32, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t1[1], MD5C33, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t3[0], MD5C34, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t0[3], MD5C35, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t2[2], MD5C36, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t0[1], MD5C37, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t2[0], MD5C38, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t3[3], MD5C39, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t1[2], MD5C3a, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t3[1], MD5C3b, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t1[0], MD5C3c, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t2[3], MD5C3d, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t0[2], MD5C3e, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t2[1], MD5C3f, MD5S33); a += MD5M_A; b += MD5M_B; @@ -145,12 +165,6 @@ static void m05100s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le const u32 gid = get_global_id (0); const u32 lid = get_local_id (0); - /** - * base - */ - - w3[2] = pw_len * 8; - /** * digest */ @@ -167,7 +181,7 @@ static void m05100s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le * loop */ - u32 w0l = w0[0]; + const u32 w0l = w0[0]; for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { @@ -175,78 +189,104 @@ static void m05100s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le const u32x w0lr = w0l | w0r; + u32x t0[4]; + u32x t1[4]; + u32x t2[4]; + u32x t3[4]; + + t0[0] = w0lr; + t0[1] = w0[1]; + t0[2] = w0[2]; + t0[3] = w0[3]; + t1[0] = w1[0]; + t1[1] = w1[1]; + t1[2] = w1[2]; + t1[3] = w1[3]; + t2[0] = w2[0]; + t2[1] = w2[1]; + t2[2] = w2[2]; + t2[3] = w2[3]; + t3[0] = w3[0]; + t3[1] = w3[1]; + t3[2] = pw_len * 8; + t3[3] = 0; + + /** + * md5 + */ + u32x a = MD5M_A; u32x b = MD5M_B; u32x c = MD5M_C; u32x d = MD5M_D; - MD5_STEP (MD5_Fo, a, b, c, d, w0lr, MD5C00, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w0[1], MD5C01, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w0[2], MD5C02, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w0[3], MD5C03, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w1[0], MD5C04, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w1[1], MD5C05, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w1[2], MD5C06, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w1[3], MD5C07, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w2[0], MD5C08, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w2[1], MD5C09, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w2[2], MD5C0a, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w2[3], MD5C0b, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w3[0], MD5C0c, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w3[1], MD5C0d, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w3[2], MD5C0e, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w3[3], MD5C0f, MD5S03); - - MD5_STEP (MD5_Go, a, b, c, d, w0[1], MD5C10, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w1[2], MD5C11, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w2[3], MD5C12, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w0lr, MD5C13, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w1[1], MD5C14, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w2[2], MD5C15, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w3[3], MD5C16, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w1[0], MD5C17, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w2[1], MD5C18, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w3[2], MD5C19, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w0[3], MD5C1a, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w2[0], MD5C1b, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w3[1], MD5C1c, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w0[2], MD5C1d, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w1[3], MD5C1e, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w3[0], MD5C1f, MD5S13); - - MD5_STEP (MD5_H , a, b, c, d, w1[1], MD5C20, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w2[0], MD5C21, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w2[3], MD5C22, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w3[2], MD5C23, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w0[1], MD5C24, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w1[0], MD5C25, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w1[3], MD5C26, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w2[2], MD5C27, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w3[1], MD5C28, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w0lr, MD5C29, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w0[3], MD5C2a, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w1[2], MD5C2b, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w2[1], MD5C2c, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w3[0], MD5C2d, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w3[3], MD5C2e, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w0[2], MD5C2f, MD5S23); - - MD5_STEP (MD5_I , a, b, c, d, w0lr, MD5C30, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w1[3], MD5C31, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w3[2], MD5C32, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w1[1], MD5C33, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w3[0], MD5C34, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w0[3], MD5C35, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w2[2], MD5C36, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w0[1], MD5C37, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w2[0], MD5C38, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w3[3], MD5C39, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w1[2], MD5C3a, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w3[1], MD5C3b, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w1[0], MD5C3c, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w2[3], MD5C3d, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w0[2], MD5C3e, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w2[1], MD5C3f, MD5S33); + MD5_STEP (MD5_Fo, a, b, c, d, t0[0], MD5C00, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t0[1], MD5C01, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t0[2], MD5C02, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t0[3], MD5C03, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t1[0], MD5C04, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t1[1], MD5C05, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t1[2], MD5C06, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t1[3], MD5C07, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t2[0], MD5C08, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t2[1], MD5C09, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t2[2], MD5C0a, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t2[3], MD5C0b, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, t3[0], MD5C0c, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, t3[1], MD5C0d, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, t3[2], MD5C0e, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, t3[3], MD5C0f, MD5S03); + + MD5_STEP (MD5_Go, a, b, c, d, t0[1], MD5C10, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t1[2], MD5C11, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t2[3], MD5C12, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t0[0], MD5C13, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t1[1], MD5C14, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t2[2], MD5C15, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t3[3], MD5C16, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t1[0], MD5C17, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t2[1], MD5C18, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t3[2], MD5C19, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t0[3], MD5C1a, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t2[0], MD5C1b, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, t3[1], MD5C1c, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, t0[2], MD5C1d, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, t1[3], MD5C1e, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, t3[0], MD5C1f, MD5S13); + + MD5_STEP (MD5_H , a, b, c, d, t1[1], MD5C20, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t2[0], MD5C21, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t2[3], MD5C22, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t3[2], MD5C23, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t0[1], MD5C24, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t1[0], MD5C25, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t1[3], MD5C26, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t2[2], MD5C27, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t3[1], MD5C28, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t0[0], MD5C29, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t0[3], MD5C2a, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t1[2], MD5C2b, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, t2[1], MD5C2c, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, t3[0], MD5C2d, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, t3[3], MD5C2e, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, t0[2], MD5C2f, MD5S23); + + MD5_STEP (MD5_I , a, b, c, d, t0[0], MD5C30, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t1[3], MD5C31, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t3[2], MD5C32, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t1[1], MD5C33, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t3[0], MD5C34, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t0[3], MD5C35, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t2[2], MD5C36, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t0[1], MD5C37, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t2[0], MD5C38, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t3[3], MD5C39, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t1[2], MD5C3a, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t3[1], MD5C3b, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, t1[0], MD5C3c, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, t2[3], MD5C3d, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, t0[2], MD5C3e, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, t2[1], MD5C3f, MD5S33); a += MD5M_A; b += MD5M_B;