From 74c16594268fe0d191a17bdefc561d1c4a5d5f64 Mon Sep 17 00:00:00 2001 From: Jens Steube Date: Thu, 4 Feb 2016 22:31:51 +0100 Subject: [PATCH] Converted to new SIMD: -m 11200 -a 0 --- OpenCL/m11200_a0.cl | 218 ++++++++++++++++---------------------------- 1 file changed, 81 insertions(+), 137 deletions(-) diff --git a/OpenCL/m11200_a0.cl b/OpenCL/m11200_a0.cl index 2161dd4..bd7248c 100644 --- a/OpenCL/m11200_a0.cl +++ b/OpenCL/m11200_a0.cl @@ -5,6 +5,8 @@ #define _SHA1_ +#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" __kernel void m11200_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) { @@ -60,47 +60,24 @@ __kernel void m11200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, u32 salt_buf[5]; - salt_buf[0] = swap32 (salt_bufs[salt_pos].salt_buf[0]); - salt_buf[1] = swap32 (salt_bufs[salt_pos].salt_buf[1]); - salt_buf[2] = swap32 (salt_bufs[salt_pos].salt_buf[2]); - salt_buf[3] = swap32 (salt_bufs[salt_pos].salt_buf[3]); - salt_buf[4] = swap32 (salt_bufs[salt_pos].salt_buf[4]); + salt_buf[0] = swap32_S (salt_bufs[salt_pos].salt_buf[0]); + salt_buf[1] = swap32_S (salt_bufs[salt_pos].salt_buf[1]); + salt_buf[2] = swap32_S (salt_bufs[salt_pos].salt_buf[2]); + salt_buf[3] = swap32_S (salt_bufs[salt_pos].salt_buf[3]); + salt_buf[4] = swap32_S (salt_bufs[salt_pos].salt_buf[4]); /** * 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]; - - 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; + u32x w0[4] = { 0 }; + u32x w1[4] = { 0 }; + u32x w2[4] = { 0 }; + u32x w3[4] = { 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); @@ -108,28 +85,28 @@ __kernel void m11200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, * sha1 ($pass) */ - u32 w0_t = swap32 (w0[0]); - u32 w1_t = swap32 (w0[1]); - u32 w2_t = swap32 (w0[2]); - u32 w3_t = swap32 (w0[3]); - u32 w4_t = swap32 (w1[0]); - u32 w5_t = swap32 (w1[1]); - u32 w6_t = swap32 (w1[2]); - u32 w7_t = swap32 (w1[3]); - u32 w8_t = swap32 (w2[0]); - u32 w9_t = swap32 (w2[1]); - u32 wa_t = swap32 (w2[2]); - u32 wb_t = swap32 (w2[3]); - u32 wc_t = swap32 (w3[0]); - u32 wd_t = swap32 (w3[1]); - u32 we_t = 0; - u32 wf_t = out_len * 8; - - u32 a = SHA1M_A; - u32 b = SHA1M_B; - u32 c = SHA1M_C; - u32 d = SHA1M_D; - u32 e = SHA1M_E; + u32x w0_t = swap32 (w0[0]); + u32x w1_t = swap32 (w0[1]); + u32x w2_t = swap32 (w0[2]); + u32x w3_t = swap32 (w0[3]); + u32x w4_t = swap32 (w1[0]); + u32x w5_t = swap32 (w1[1]); + u32x w6_t = swap32 (w1[2]); + u32x w7_t = swap32 (w1[3]); + u32x w8_t = swap32 (w2[0]); + u32x w9_t = swap32 (w2[1]); + u32x wa_t = swap32 (w2[2]); + u32x wb_t = swap32 (w2[3]); + u32x wc_t = swap32 (w3[0]); + u32x wd_t = swap32 (w3[1]); + u32x we_t = 0; + u32x wf_t = out_len * 8; + + u32x a = SHA1M_A; + u32x b = SHA1M_B; + u32x c = SHA1M_C; + u32x d = SHA1M_D; + u32x e = SHA1M_E; #undef K #define K SHA1C00 @@ -227,11 +204,11 @@ __kernel void m11200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, we_t = rotl32 ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP (SHA1_F1, c, d, e, a, b, we_t); wf_t = rotl32 ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP (SHA1_F1, b, c, d, e, a, wf_t); - u32 plain_sha1_a = a + SHA1M_A; - u32 plain_sha1_b = b + SHA1M_B; - u32 plain_sha1_c = c + SHA1M_C; - u32 plain_sha1_d = d + SHA1M_D; - u32 plain_sha1_e = e + SHA1M_E; + u32x plain_sha1_a = a + SHA1M_A; + u32x plain_sha1_b = b + SHA1M_B; + u32x plain_sha1_c = c + SHA1M_C; + u32x plain_sha1_d = d + SHA1M_D; + u32x plain_sha1_e = e + SHA1M_E; /** * sha1 (sha1 ($pass)) @@ -498,12 +475,7 @@ __kernel void m11200_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, d ^= plain_sha1_d; e ^= plain_sha1_e; - const u32 r0 = d; - const u32 r1 = e; - const u32 r2 = c; - const u32 r3 = b; - - #include COMPARE_M + COMPARE_M_SIMD (d, e, c, b); } } @@ -565,47 +537,24 @@ __kernel void m11200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, u32 salt_buf[5]; - salt_buf[0] = swap32 (salt_bufs[salt_pos].salt_buf[0]); - salt_buf[1] = swap32 (salt_bufs[salt_pos].salt_buf[1]); - salt_buf[2] = swap32 (salt_bufs[salt_pos].salt_buf[2]); - salt_buf[3] = swap32 (salt_bufs[salt_pos].salt_buf[3]); - salt_buf[4] = swap32 (salt_bufs[salt_pos].salt_buf[4]); + salt_buf[0] = swap32_S (salt_bufs[salt_pos].salt_buf[0]); + salt_buf[1] = swap32_S (salt_bufs[salt_pos].salt_buf[1]); + salt_buf[2] = swap32_S (salt_bufs[salt_pos].salt_buf[2]); + salt_buf[3] = swap32_S (salt_bufs[salt_pos].salt_buf[3]); + salt_buf[4] = swap32_S (salt_bufs[salt_pos].salt_buf[4]); /** * 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]; - - 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]; + u32x w0[4] = { 0 }; + u32x w1[4] = { 0 }; + u32x w2[4] = { 0 }; + u32x w3[4] = { 0 }; - 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); @@ -613,28 +562,28 @@ __kernel void m11200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, * sha1 ($pass) */ - u32 w0_t = swap32 (w0[0]); - u32 w1_t = swap32 (w0[1]); - u32 w2_t = swap32 (w0[2]); - u32 w3_t = swap32 (w0[3]); - u32 w4_t = swap32 (w1[0]); - u32 w5_t = swap32 (w1[1]); - u32 w6_t = swap32 (w1[2]); - u32 w7_t = swap32 (w1[3]); - u32 w8_t = swap32 (w2[0]); - u32 w9_t = swap32 (w2[1]); - u32 wa_t = swap32 (w2[2]); - u32 wb_t = swap32 (w2[3]); - u32 wc_t = swap32 (w3[0]); - u32 wd_t = swap32 (w3[1]); - u32 we_t = 0; - u32 wf_t = out_len * 8; - - u32 a = SHA1M_A; - u32 b = SHA1M_B; - u32 c = SHA1M_C; - u32 d = SHA1M_D; - u32 e = SHA1M_E; + u32x w0_t = swap32 (w0[0]); + u32x w1_t = swap32 (w0[1]); + u32x w2_t = swap32 (w0[2]); + u32x w3_t = swap32 (w0[3]); + u32x w4_t = swap32 (w1[0]); + u32x w5_t = swap32 (w1[1]); + u32x w6_t = swap32 (w1[2]); + u32x w7_t = swap32 (w1[3]); + u32x w8_t = swap32 (w2[0]); + u32x w9_t = swap32 (w2[1]); + u32x wa_t = swap32 (w2[2]); + u32x wb_t = swap32 (w2[3]); + u32x wc_t = swap32 (w3[0]); + u32x wd_t = swap32 (w3[1]); + u32x we_t = 0; + u32x wf_t = out_len * 8; + + u32x a = SHA1M_A; + u32x b = SHA1M_B; + u32x c = SHA1M_C; + u32x d = SHA1M_D; + u32x e = SHA1M_E; #undef K #define K SHA1C00 @@ -732,11 +681,11 @@ __kernel void m11200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, we_t = rotl32 ((wb_t ^ w6_t ^ w0_t ^ we_t), 1u); SHA1_STEP (SHA1_F1, c, d, e, a, b, we_t); wf_t = rotl32 ((wc_t ^ w7_t ^ w1_t ^ wf_t), 1u); SHA1_STEP (SHA1_F1, b, c, d, e, a, wf_t); - u32 plain_sha1_a = a + SHA1M_A; - u32 plain_sha1_b = b + SHA1M_B; - u32 plain_sha1_c = c + SHA1M_C; - u32 plain_sha1_d = d + SHA1M_D; - u32 plain_sha1_e = e + SHA1M_E; + u32x plain_sha1_a = a + SHA1M_A; + u32x plain_sha1_b = b + SHA1M_B; + u32x plain_sha1_c = c + SHA1M_C; + u32x plain_sha1_d = d + SHA1M_D; + u32x plain_sha1_e = e + SHA1M_E; /** * sha1 (sha1 ($pass)) @@ -1003,12 +952,7 @@ __kernel void m11200_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, d ^= plain_sha1_d; e ^= plain_sha1_e; - const u32 r0 = d; - const u32 r1 = e; - const u32 r2 = c; - const u32 r3 = b; - - #include COMPARE_S + COMPARE_S_SIMD (d, e, c, b); } } -- 2.43.0