From: Jens Steube Date: Fri, 5 Feb 2016 20:44:23 +0000 (+0100) Subject: Converted to new SIMD: -m 3100 -a 0 X-Git-Tag: v3.00-beta~345 X-Git-Url: https://www.flypig.org.uk/git/?a=commitdiff_plain;h=ceb96a773536b060b9cf2cc4ef6f026a5cb625b8;p=hashcat.git Converted to new SIMD: -m 3100 -a 0 --- diff --git a/OpenCL/m03100_a0.cl b/OpenCL/m03100_a0.cl index 5dc106e..cdcdf09 100644 --- a/OpenCL/m03100_a0.cl +++ b/OpenCL/m03100_a0.cl @@ -7,6 +7,8 @@ #define _DES_ +#define NEW_SIMD_CODE + #include "include/constants.h" #include "include/kernel_vendor.h" @@ -20,9 +22,7 @@ #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" #define PERM_OP(a,b,tt,n,m) \ { \ @@ -368,12 +368,12 @@ __constant u32 c_skb[8][64] = #define BOX(i,n,S) (u32x) ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7]) #endif -static void _des_crypt_encrypt (u32 iv[2], u32 data[2], u32 Kc[16], u32 Kd[16], __local u32 (*s_SPtrans)[64]) +static void _des_crypt_encrypt (u32x iv[2], u32x data[2], u32x Kc[16], u32x Kd[16], __local u32 (*s_SPtrans)[64]) { - u32 tt; + u32x tt; - u32 r = data[0]; - u32 l = data[1]; + u32x r = data[0]; + u32x l = data[1]; IP (r, l, tt); @@ -383,8 +383,8 @@ static void _des_crypt_encrypt (u32 iv[2], u32 data[2], u32 Kc[16], u32 Kd[16], #pragma unroll 16 for (u32 i = 0; i < 16; i += 2) { - u32 u; - u32 t; + u32x u; + u32x t; u = Kc[i + 0] ^ r; t = Kd[i + 0] ^ rotl32 (r, 28u); @@ -420,9 +420,9 @@ static void _des_crypt_encrypt (u32 iv[2], u32 data[2], u32 Kc[16], u32 Kd[16], iv[1] = r; } -static void _des_crypt_keysetup (u32 c, u32 d, u32 Kc[16], u32 Kd[16], __local u32 (*s_skb)[64]) +static void _des_crypt_keysetup (u32x c, u32x d, u32x Kc[16], u32x Kd[16], __local u32 (*s_skb)[64]) { - u32 tt; + u32x tt; PERM_OP (d, c, tt, 4, 0x0f0f0f0f); HPERM_OP (c, tt, 2, 0xcccc0000); @@ -455,32 +455,32 @@ static void _des_crypt_keysetup (u32 c, u32 d, u32 Kc[16], u32 Kd[16], __local u c = c & 0x0fffffff; d = d & 0x0fffffff; - const u32 c00 = (c >> 0) & 0x0000003f; - const u32 c06 = (c >> 6) & 0x00383003; - const u32 c07 = (c >> 7) & 0x0000003c; - const u32 c13 = (c >> 13) & 0x0000060f; - const u32 c20 = (c >> 20) & 0x00000001; - - u32 s = BOX (((c00 >> 0) & 0xff), 0, s_skb) - | BOX (((c06 >> 0) & 0xff) - |((c07 >> 0) & 0xff), 1, s_skb) - | BOX (((c13 >> 0) & 0xff) - |((c06 >> 8) & 0xff), 2, s_skb) - | BOX (((c20 >> 0) & 0xff) - |((c13 >> 8) & 0xff) - |((c06 >> 16) & 0xff), 3, s_skb); - - const u32 d00 = (d >> 0) & 0x00003c3f; - const u32 d07 = (d >> 7) & 0x00003f03; - const u32 d21 = (d >> 21) & 0x0000000f; - const u32 d22 = (d >> 22) & 0x00000030; - - u32 t = BOX (((d00 >> 0) & 0xff), 4, s_skb) - | BOX (((d07 >> 0) & 0xff) - |((d00 >> 8) & 0xff), 5, s_skb) - | BOX (((d07 >> 8) & 0xff), 6, s_skb) - | BOX (((d21 >> 0) & 0xff) - |((d22 >> 0) & 0xff), 7, s_skb); + const u32x c00 = (c >> 0) & 0x0000003f; + const u32x c06 = (c >> 6) & 0x00383003; + const u32x c07 = (c >> 7) & 0x0000003c; + const u32x c13 = (c >> 13) & 0x0000060f; + const u32x c20 = (c >> 20) & 0x00000001; + + u32x s = BOX (((c00 >> 0) & 0xff), 0, s_skb) + | BOX (((c06 >> 0) & 0xff) + |((c07 >> 0) & 0xff), 1, s_skb) + | BOX (((c13 >> 0) & 0xff) + |((c06 >> 8) & 0xff), 2, s_skb) + | BOX (((c20 >> 0) & 0xff) + |((c13 >> 8) & 0xff) + |((c06 >> 16) & 0xff), 3, s_skb); + + const u32x d00 = (d >> 0) & 0x00003c3f; + const u32x d07 = (d >> 7) & 0x00003f03; + const u32x d21 = (d >> 21) & 0x0000000f; + const u32x d22 = (d >> 22) & 0x00000030; + + u32x t = BOX (((d00 >> 0) & 0xff), 4, s_skb) + | BOX (((d07 >> 0) & 0xff) + |((d00 >> 8) & 0xff), 5, s_skb) + | BOX (((d07 >> 8) & 0xff), 6, s_skb) + | BOX (((d21 >> 0) & 0xff) + |((d22 >> 0) & 0xff), 7, s_skb); Kc[i] = ((t << 16) | (s & 0x0000ffff)); Kd[i] = ((s >> 16) | (t & 0xffff0000)); @@ -576,37 +576,14 @@ __kernel void m03100_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 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]; + u32x w0[4] = { 0 }; + u32x w1[4] = { 0 }; + u32x w2[4] = { 0 }; + u32x w3[4] = { 0 }; - 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); const u32 salt_word_len = (salt_len + out_len) * 2; @@ -614,28 +591,28 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * prepend salt */ - u32 w0_t[4]; + u32x w0_t[4]; w0_t[0] = w0[0]; w0_t[1] = w0[1]; w0_t[2] = w0[2]; w0_t[3] = w0[3]; - u32 w1_t[4]; + u32x w1_t[4]; w1_t[0] = w1[0]; w1_t[1] = w1[1]; w1_t[2] = w1[2]; w1_t[3] = w1[3]; - u32 w2_t[4]; + u32x w2_t[4]; w2_t[0] = w2[0]; w2_t[1] = w2[1]; w2_t[2] = w2[2]; w2_t[3] = w2[3]; - u32 w3_t[4]; + u32x w3_t[4]; w3_t[0] = w3[0]; w3_t[1] = w3[1]; @@ -653,7 +630,7 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu w1_t[2] |= salt_buf1[2]; w1_t[3] |= salt_buf1[3]; - u32 dst[16]; + u32x dst[16]; dst[ 0] = w0_t[0]; dst[ 1] = w0_t[1]; @@ -677,7 +654,7 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * plus LEFT_ROTATE by 2 */ - u32 Kc[16]; + u32x Kc[16]; Kc[ 0] = 0x64649040; Kc[ 1] = 0x14909858; @@ -696,7 +673,7 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu Kc[14] = 0x584020b4; Kc[15] = 0x00742c4c; - u32 Kd[16]; + u32x Kd[16]; Kd[ 0] = 0xa42ce40c; Kd[ 1] = 0x64689858; @@ -719,14 +696,14 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * key1 (generate key) */ - u32 iv[2]; + u32x iv[2]; iv[0] = 0; iv[1] = 0; for (u32 j = 0, k = 0; j < salt_word_len; j += 8, k++) { - u32 data[2]; + u32x data[2]; data[0] = ((dst[k] << 16) & 0xff000000) | ((dst[k] << 8) & 0x0000ff00); data[1] = ((dst[k] >> 0) & 0xff000000) | ((dst[k] >> 8) & 0x0000ff00); @@ -748,7 +725,7 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu for (u32 j = 0, k = 0; j < salt_word_len; j += 8, k++) { - u32 data[2]; + u32x data[2]; data[0] = ((dst[k] << 16) & 0xff000000) | ((dst[k] << 8) & 0x0000ff00); data[1] = ((dst[k] >> 0) & 0xff000000) | ((dst[k] >> 8) & 0x0000ff00); @@ -763,12 +740,12 @@ __kernel void m03100_m04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * cmp */ - const u32 r0 = iv[0]; - const u32 r1 = iv[1]; - const u32 r2 = 0; - const u32 r3 = 0; + const u32x r0 = iv[0]; + const u32x r1 = iv[1]; + const u32x r2 = 0; + const u32x r3 = 0; - #include COMPARE_M + COMPARE_M_SIMD (r0, r1, r2, r3); } } @@ -878,37 +855,14 @@ __kernel void m03100_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 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); const u32 salt_word_len = (salt_len + out_len) * 2; @@ -916,28 +870,28 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * prepend salt */ - u32 w0_t[4]; + u32x w0_t[4]; w0_t[0] = w0[0]; w0_t[1] = w0[1]; w0_t[2] = w0[2]; w0_t[3] = w0[3]; - u32 w1_t[4]; + u32x w1_t[4]; w1_t[0] = w1[0]; w1_t[1] = w1[1]; w1_t[2] = w1[2]; w1_t[3] = w1[3]; - u32 w2_t[4]; + u32x w2_t[4]; w2_t[0] = w2[0]; w2_t[1] = w2[1]; w2_t[2] = w2[2]; w2_t[3] = w2[3]; - u32 w3_t[4]; + u32x w3_t[4]; w3_t[0] = w3[0]; w3_t[1] = w3[1]; @@ -955,7 +909,7 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu w1_t[2] |= salt_buf1[2]; w1_t[3] |= salt_buf1[3]; - u32 dst[16]; + u32x dst[16]; dst[ 0] = w0_t[0]; dst[ 1] = w0_t[1]; @@ -979,7 +933,7 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * plus LEFT_ROTATE by 2 */ - u32 Kc[16]; + u32x Kc[16]; Kc[ 0] = 0x64649040; Kc[ 1] = 0x14909858; @@ -998,7 +952,7 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu Kc[14] = 0x584020b4; Kc[15] = 0x00742c4c; - u32 Kd[16]; + u32x Kd[16]; Kd[ 0] = 0xa42ce40c; Kd[ 1] = 0x64689858; @@ -1021,14 +975,14 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * key1 (generate key) */ - u32 iv[2]; + u32x iv[2]; iv[0] = 0; iv[1] = 0; for (u32 j = 0, k = 0; j < salt_word_len; j += 8, k++) { - u32 data[2]; + u32x data[2]; data[0] = ((dst[k] << 16) & 0xff000000) | ((dst[k] << 8) & 0x0000ff00); data[1] = ((dst[k] >> 0) & 0xff000000) | ((dst[k] >> 8) & 0x0000ff00); @@ -1050,7 +1004,7 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu for (u32 j = 0, k = 0; j < salt_word_len; j += 8, k++) { - u32 data[2]; + u32x data[2]; data[0] = ((dst[k] << 16) & 0xff000000) | ((dst[k] << 8) & 0x0000ff00); data[1] = ((dst[k] >> 0) & 0xff000000) | ((dst[k] >> 8) & 0x0000ff00); @@ -1065,12 +1019,12 @@ __kernel void m03100_s04 (__global pw_t *pws, __global kernel_rule_t * rules_bu * cmp */ - const u32 r0 = iv[0]; - const u32 r1 = iv[1]; - const u32 r2 = 0; - const u32 r3 = 0; + const u32x r0 = iv[0]; + const u32x r1 = iv[1]; + const u32x r2 = 0; + const u32x r3 = 0; - #include COMPARE_S + COMPARE_M_SIMD (r0, r1, r2, r3); } }