From 737678284f3bbf5a4d0639c9697d114476e51819 Mon Sep 17 00:00:00 2001 From: Jens Steube Date: Thu, 4 Feb 2016 23:12:24 +0100 Subject: [PATCH] Converted to new SIMD: -m 10100 -a 0 --- OpenCL/m10100_a0.cl | 176 ++++++++++++++++++++++---------------------- OpenCL/m10400_a0.cl | 4 +- OpenCL/m11100_a0.cl | 4 +- OpenCL/m11500_a0.cl | 4 +- 4 files changed, 93 insertions(+), 95 deletions(-) diff --git a/OpenCL/m10100_a0.cl b/OpenCL/m10100_a0.cl index a6373827a..6bbd6ad07 100644 --- a/OpenCL/m10100_a0.cl +++ b/OpenCL/m10100_a0.cl @@ -5,6 +5,8 @@ #define _SIPHASH_ +#define NEW_SIMD_CODE + #include "include/constants.h" #include "include/kernel_vendor.h" @@ -18,15 +20,13 @@ #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" #define SIPROUND(v0,v1,v2,v3) \ (v0) += (v1); \ (v1) = rotl64 ((v1), 13); \ (v1) ^= (v0); \ - (v0) = as_ulong (as_uint2 ((v0)).s10); \ + (v0) = rotl64 ((v0), 32); \ (v2) += (v3); \ (v3) = rotl64 ((v3), 16); \ (v3) ^= (v2); \ @@ -36,7 +36,7 @@ (v2) += (v1); \ (v1) = rotl64 ((v1), 17); \ (v1) ^= (v2); \ - (v2) = as_ulong (as_uint2 ((v2)).s10); + (v2) = rotl64 ((v2), 32) __kernel void m10100_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) { @@ -74,10 +74,10 @@ __kernel void m10100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * base */ - u64 v0p = SIPHASHM_0; - u64 v1p = SIPHASHM_1; - u64 v2p = SIPHASHM_2; - u64 v3p = SIPHASHM_3; + u64x v0p = SIPHASHM_0; + u64x v1p = SIPHASHM_1; + u64x v2p = SIPHASHM_2; + u64x v3p = SIPHASHM_3; v0p ^= hl32_to_64 (salt_bufs[salt_pos].salt_buf[1], salt_bufs[salt_pos].salt_buf[0]); v1p ^= hl32_to_64 (salt_bufs[salt_pos].salt_buf[3], salt_bufs[salt_pos].salt_buf[2]); @@ -88,44 +88,46 @@ __kernel void m10100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * 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 w[16]; + u32x w0[4] = { 0 }; + u32x w1[4] = { 0 }; + u32x w2[4] = { 0 }; + u32x w3[4] = { 0 }; - w[ 0] = pw_buf0[0]; - w[ 1] = pw_buf0[1]; - w[ 2] = pw_buf0[2]; - w[ 3] = pw_buf0[3]; - w[ 4] = pw_buf1[0]; - w[ 5] = pw_buf1[1]; - w[ 6] = pw_buf1[2]; - w[ 7] = pw_buf1[3]; - w[ 8] = 0; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = 0; + const u32 out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1); - const u32 out_len = apply_rules (rules_buf[il_pos].cmds, &w[0], &w[4], pw_len); + switch (out_len / 8) + { + case 0: w0[1] |= out_len << 24; break; + case 1: w0[3] |= out_len << 24; break; + case 2: w1[1] |= out_len << 24; break; + case 3: w1[3] |= out_len << 24; break; + } - u64 *w_ptr = (u64 *) w; - - w_ptr[out_len / 8] |= (u64) out_len << 56; - - u64 v0 = v0p; - u64 v1 = v1p; - u64 v2 = v2p; - u64 v3 = v3p; + u64x v0 = v0p; + u64x v1 = v1p; + u64x v2 = v2p; + u64x v3 = v3p; int i; int j; - for (i = 0, j = 0; i <= pw_len; i += 8, j += 2) + for (i = 0, j = 0; i <= out_len && i < 16; i += 8, j += 2) { - u64 m = hl32_to_64 (w[j + 1], w[j + 0]); + u64x m = hl32_to_64 (w0[j + 1], w0[j + 0]); + + v3 ^= m; + + SIPROUND (v0, v1, v2, v3); + SIPROUND (v0, v1, v2, v3); + + v0 ^= m; + } + + for ( j = 0; i <= out_len && i < 32; i += 8, j += 2) + { + u64x m = hl32_to_64 (w1[j + 1], w1[j + 0]); v3 ^= m; @@ -142,17 +144,14 @@ __kernel void m10100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu SIPROUND (v0, v1, v2, v3); SIPROUND (v0, v1, v2, v3); - const u64 v = v0 ^ v1 ^ v2 ^ v3; + const u64x v = v0 ^ v1 ^ v2 ^ v3; - const u32 a = l32_from_64 (v); - const u32 b = h32_from_64 (v); + const u32x a = l32_from_64 (v); + const u32x b = h32_from_64 (v); + const u32x c = 0; + const u32x d = 0; - 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); } } @@ -200,10 +199,10 @@ __kernel void m10100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * base */ - u64 v0p = SIPHASHM_0; - u64 v1p = SIPHASHM_1; - u64 v2p = SIPHASHM_2; - u64 v3p = SIPHASHM_3; + u64x v0p = SIPHASHM_0; + u64x v1p = SIPHASHM_1; + u64x v2p = SIPHASHM_2; + u64x v3p = SIPHASHM_3; v0p ^= hl32_to_64 (salt_bufs[salt_pos].salt_buf[1], salt_bufs[salt_pos].salt_buf[0]); v1p ^= hl32_to_64 (salt_bufs[salt_pos].salt_buf[3], salt_bufs[salt_pos].salt_buf[2]); @@ -226,44 +225,46 @@ __kernel void m10100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * 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 w[16]; + u32x w0[4] = { 0 }; + u32x w1[4] = { 0 }; + u32x w2[4] = { 0 }; + u32x w3[4] = { 0 }; - w[ 0] = pw_buf0[0]; - w[ 1] = pw_buf0[1]; - w[ 2] = pw_buf0[2]; - w[ 3] = pw_buf0[3]; - w[ 4] = pw_buf1[0]; - w[ 5] = pw_buf1[1]; - w[ 6] = pw_buf1[2]; - w[ 7] = pw_buf1[3]; - w[ 8] = 0; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = 0; + const u32 out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1); - const u32 out_len = apply_rules (rules_buf[il_pos].cmds, &w[0], &w[4], pw_len); + switch (out_len / 8) + { + case 0: w0[1] |= out_len << 24; break; + case 1: w0[3] |= out_len << 24; break; + case 2: w1[1] |= out_len << 24; break; + case 3: w1[3] |= out_len << 24; break; + } - u64 *w_ptr = (u64 *) w; - - w_ptr[out_len / 8] |= (u64) out_len << 56; - - u64 v0 = v0p; - u64 v1 = v1p; - u64 v2 = v2p; - u64 v3 = v3p; + u64x v0 = v0p; + u64x v1 = v1p; + u64x v2 = v2p; + u64x v3 = v3p; int i; int j; - for (i = 0, j = 0; i <= pw_len; i += 8, j += 2) + for (i = 0, j = 0; i <= out_len && i < 16; i += 8, j += 2) { - u64 m = hl32_to_64 (w[j + 1], w[j + 0]); + u64x m = hl32_to_64 (w0[j + 1], w0[j + 0]); + + v3 ^= m; + + SIPROUND (v0, v1, v2, v3); + SIPROUND (v0, v1, v2, v3); + + v0 ^= m; + } + + for ( j = 0; i <= out_len && i < 32; i += 8, j += 2) + { + u64x m = hl32_to_64 (w1[j + 1], w1[j + 0]); v3 ^= m; @@ -280,17 +281,14 @@ __kernel void m10100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu SIPROUND (v0, v1, v2, v3); SIPROUND (v0, v1, v2, v3); - const u64 v = v0 ^ v1 ^ v2 ^ v3; + const u64x v = v0 ^ v1 ^ v2 ^ v3; - const u32 a = l32_from_64 (v); - const u32 b = h32_from_64 (v); + const u32x a = l32_from_64 (v); + const u32x b = h32_from_64 (v); + const u32x c = 0; + const u32x d = 0; - 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/m10400_a0.cl b/OpenCL/m10400_a0.cl index 6804cbca5..7f468f54c 100644 --- a/OpenCL/m10400_a0.cl +++ b/OpenCL/m10400_a0.cl @@ -360,7 +360,7 @@ __kernel void m10400_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, w3_t[2] = 0; w3_t[3] = 0; - switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, pw_len); + switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, out_len); // add password // truncate at 32 is wanted, not a bug! @@ -579,7 +579,7 @@ __kernel void m10400_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, w3_t[2] = 0; w3_t[3] = 0; - switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, pw_len); + switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, out_len); // add password // truncate at 32 is wanted, not a bug! diff --git a/OpenCL/m11100_a0.cl b/OpenCL/m11100_a0.cl index fbc1b8171..e4c38c942 100644 --- a/OpenCL/m11100_a0.cl +++ b/OpenCL/m11100_a0.cl @@ -154,7 +154,7 @@ __kernel void m11100_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, * append the salt */ - switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, pw_len); + switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, out_len); const u32 pw_salt_len = out_len + salt_len; @@ -512,7 +512,7 @@ __kernel void m11100_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, * append the salt */ - switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, pw_len); + switch_buffer_by_offset_le (w0_t, w1_t, w2_t, w3_t, out_len); const u32 pw_salt_len = out_len + salt_len; diff --git a/OpenCL/m11500_a0.cl b/OpenCL/m11500_a0.cl index 58aac5468..9b6d1bd40 100644 --- a/OpenCL/m11500_a0.cl +++ b/OpenCL/m11500_a0.cl @@ -201,7 +201,7 @@ __kernel void m11500_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, w[14] = 0; w[15] = 0; - u32x a = crc32 (w, pw_len, iv); + u32x a = crc32 (w, out_len, iv); u32x b = 0; u32x c = 0; u32x d = 0; @@ -296,7 +296,7 @@ __kernel void m11500_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, w[14] = 0; w[15] = 0; - u32x a = crc32 (w, pw_len, iv); + u32x a = crc32 (w, out_len, iv); u32x b = 0; u32x c = 0; u32x d = 0;