Permalink
Please
sign in to comment.
Showing
with
2,358 additions
and 2 deletions.
- +382 −0 OpenCL/m22200_a0-optimized.cl
- +133 −0 OpenCL/m22200_a0-pure.cl
- +511 −0 OpenCL/m22200_a1-optimized.cl
- +124 −0 OpenCL/m22200_a1-pure.cl
- +735 −0 OpenCL/m22200_a3-optimized.cl
- +157 −0 OpenCL/m22200_a3-pure.cl
- +1 −0 docs/changes.txt
- +2 −1 docs/readme.txt
- +1 −1 src/modules/module_08100.c
- +261 −0 src/modules/module_22200.c
- +51 −0 tools/test_modules/m22200.pm
| @@ -0,0 +1,382 @@ | ||
| /** | ||
| * Author......: See docs/credits.txt | ||
| * License.....: MIT | ||
| */ | ||
|
|
||
| #define NEW_SIMD_CODE | ||
|
|
||
| #ifdef KERNEL_STATIC | ||
| #include "inc_vendor.h" | ||
| #include "inc_types.h" | ||
| #include "inc_platform.cl" | ||
| #include "inc_common.cl" | ||
| #include "inc_rp_optimized.h" | ||
| #include "inc_rp_optimized.cl" | ||
| #include "inc_simd.cl" | ||
| #include "inc_hash_sha512.cl" | ||
| #endif | ||
|
|
||
| DECLSPEC void sha512_transform_intern (const u32x *w0, const u32x *w1, const u32x *w2, const u32x *w3, u64x *digest) | ||
| { | ||
| u64x w0_t = hl32_to_64 (w0[0], w0[1]); | ||
| u64x w1_t = hl32_to_64 (w0[2], w0[3]); | ||
| u64x w2_t = hl32_to_64 (w1[0], w1[1]); | ||
| u64x w3_t = hl32_to_64 (w1[2], w1[3]); | ||
| u64x w4_t = hl32_to_64 (w2[0], w2[1]); | ||
| u64x w5_t = hl32_to_64 (w2[2], w2[3]); | ||
| u64x w6_t = hl32_to_64 (w3[0], w3[1]); | ||
| u64x w7_t = 0; | ||
| u64x w8_t = 0; | ||
| u64x w9_t = 0; | ||
| u64x wa_t = 0; | ||
| u64x wb_t = 0; | ||
| u64x wc_t = 0; | ||
| u64x wd_t = 0; | ||
| u64x we_t = 0; | ||
| u64x wf_t = hl32_to_64 (w3[2], w3[3]); | ||
|
|
||
| u64x a = digest[0]; | ||
| u64x b = digest[1]; | ||
| u64x c = digest[2]; | ||
| u64x d = digest[3]; | ||
| u64x e = digest[4]; | ||
| u64x f = digest[5]; | ||
| u64x g = digest[6]; | ||
| u64x h = digest[7]; | ||
|
|
||
| #define ROUND_EXPAND() \ | ||
| { \ | ||
| w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \ | ||
| w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \ | ||
| w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \ | ||
| w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \ | ||
| w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \ | ||
| w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \ | ||
| w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \ | ||
| w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \ | ||
| w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \ | ||
| w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \ | ||
| wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \ | ||
| wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \ | ||
| wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \ | ||
| wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \ | ||
| we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \ | ||
| wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \ | ||
| } | ||
|
|
||
| #define ROUND_STEP(i) \ | ||
| { \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \ | ||
| SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \ | ||
| } | ||
|
|
||
| ROUND_STEP (0); | ||
|
|
||
| #ifdef IS_CUDA | ||
| ROUND_EXPAND (); ROUND_STEP (16); | ||
| ROUND_EXPAND (); ROUND_STEP (32); | ||
| ROUND_EXPAND (); ROUND_STEP (48); | ||
| ROUND_EXPAND (); ROUND_STEP (64); | ||
| #else | ||
| #ifdef _unroll | ||
| #pragma unroll | ||
| #endif | ||
| for (int i = 16; i < 80; i += 16) | ||
| { | ||
| ROUND_EXPAND (); ROUND_STEP (i); | ||
| } | ||
| #endif | ||
|
|
||
| /* rev | ||
| digest[0] += a; | ||
| digest[1] += b; | ||
| digest[2] += c; | ||
| digest[3] += d; | ||
| digest[4] += e; | ||
| digest[5] += f; | ||
| digest[6] += g; | ||
| digest[7] += h; | ||
| */ | ||
|
|
||
| digest[0] = a; | ||
| digest[1] = b; | ||
| digest[2] = c; | ||
| digest[3] = d; | ||
| digest[4] = e; | ||
| digest[5] = f; | ||
| digest[6] = g; | ||
| digest[7] = h; | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_m04 (KERN_ATTR_RULES ()) | ||
| { | ||
| /** | ||
| * modifier | ||
| */ | ||
|
|
||
| const u64 lid = get_local_id (0); | ||
|
|
||
| /** | ||
| * base | ||
| */ | ||
|
|
||
| const u64 gid = get_global_id (0); | ||
|
|
||
| if (gid >= gid_max) return; | ||
|
|
||
| u32 pw_buf0[4]; | ||
| u32 pw_buf1[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]; | ||
| 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 & 63; | ||
|
|
||
| /** | ||
| * salt | ||
| */ | ||
|
|
||
| u32 salt_buf0[2]; | ||
|
|
||
| salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0]; | ||
| salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1]; | ||
|
|
||
| const u32 salt_len = salt_bufs[salt_pos].salt_len; | ||
|
|
||
| /** | ||
| * loop | ||
| */ | ||
|
|
||
| for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) | ||
| { | ||
| u32x w0[4] = { 0 }; | ||
| u32x w1[4] = { 0 }; | ||
| u32x w2[4] = { 0 }; | ||
| u32x w3[4] = { 0 }; | ||
|
|
||
| const u32x out_len = apply_rules_vect_optimized (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1); | ||
|
|
||
| append_0x80_4x4_VV (w0, w1, w2, w3, out_len + 1); | ||
|
|
||
| w0[0] = hc_swap32 (w0[0]); | ||
| w0[1] = hc_swap32 (w0[1]); | ||
| w0[2] = hc_swap32 (w0[2]); | ||
| w0[3] = hc_swap32 (w0[3]); | ||
| w1[0] = hc_swap32 (w1[0]); | ||
| w1[1] = hc_swap32 (w1[1]); | ||
| w1[2] = hc_swap32 (w1[2]); | ||
| w1[3] = hc_swap32 (w1[3]); | ||
| w2[0] = hc_swap32 (w2[0]); | ||
|
|
||
| /** | ||
| * prepend salt | ||
| */ | ||
|
|
||
| const u32x out_salt_len = out_len + salt_len; | ||
|
|
||
| u32x w0_t[4]; | ||
| u32x w1_t[4]; | ||
| u32x w2_t[4]; | ||
| u32x w3_t[4]; | ||
|
|
||
| w0_t[0] = salt_buf0[0]; | ||
| w0_t[1] = salt_buf0[1]; | ||
| w0_t[2] = w0[0]; | ||
| w0_t[3] = w0[1]; | ||
| w1_t[0] = w0[2]; | ||
| w1_t[1] = w0[3]; | ||
| w1_t[2] = w1[0]; | ||
| w1_t[3] = w1[1]; | ||
| w2_t[0] = w1[2]; | ||
| w2_t[1] = w1[3]; | ||
| w2_t[2] = w2[0]; | ||
| w2_t[3] = 0; | ||
| w3_t[0] = 0; | ||
| w3_t[1] = 0; | ||
| w3_t[2] = 0; | ||
| w3_t[3] = (out_salt_len + 1) * 8; | ||
| u64x digest[8]; | ||
|
|
||
| digest[0] = SHA512M_A; | ||
| digest[1] = SHA512M_B; | ||
| digest[2] = SHA512M_C; | ||
| digest[3] = SHA512M_D; | ||
| digest[4] = SHA512M_E; | ||
| digest[5] = SHA512M_F; | ||
| digest[6] = SHA512M_G; | ||
| digest[7] = SHA512M_H; | ||
|
|
||
| sha512_transform_intern (w0_t, w1_t, w2_t, w3_t, digest); | ||
|
|
||
| const u32x r0 = l32_from_64 (digest[7]); | ||
| const u32x r1 = h32_from_64 (digest[7]); | ||
| const u32x r2 = l32_from_64 (digest[3]); | ||
| const u32x r3 = h32_from_64 (digest[3]); | ||
|
|
||
| COMPARE_M_SIMD (r0, r1, r2, r3); | ||
| } | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_m08 (KERN_ATTR_RULES ()) | ||
| { | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_m16 (KERN_ATTR_RULES ()) | ||
| { | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_s04 (KERN_ATTR_RULES ()) | ||
| { | ||
| /** | ||
| * modifier | ||
| */ | ||
|
|
||
| const u64 lid = get_local_id (0); | ||
|
|
||
| /** | ||
| * base | ||
| */ | ||
|
|
||
| const u64 gid = get_global_id (0); | ||
|
|
||
| if (gid >= gid_max) return; | ||
|
|
||
| u32 pw_buf0[4]; | ||
| u32 pw_buf1[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]; | ||
| 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 & 63; | ||
|
|
||
| /** | ||
| * salt | ||
| */ | ||
|
|
||
| u32 salt_buf0[2]; | ||
|
|
||
| salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0]; | ||
| salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1]; | ||
|
|
||
| const u32 salt_len = salt_bufs[salt_pos].salt_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 | ||
| */ | ||
|
|
||
| for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) | ||
| { | ||
| u32x w0[4] = { 0 }; | ||
| u32x w1[4] = { 0 }; | ||
| u32x w2[4] = { 0 }; | ||
| u32x w3[4] = { 0 }; | ||
|
|
||
| const u32x out_len = apply_rules_vect_optimized (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1); | ||
|
|
||
| append_0x80_4x4_VV (w0, w1, w2, w3, out_len + 1); | ||
|
|
||
| w0[0] = hc_swap32 (w0[0]); | ||
| w0[1] = hc_swap32 (w0[1]); | ||
| w0[2] = hc_swap32 (w0[2]); | ||
| w0[3] = hc_swap32 (w0[3]); | ||
| w1[0] = hc_swap32 (w1[0]); | ||
| w1[1] = hc_swap32 (w1[1]); | ||
| w1[2] = hc_swap32 (w1[2]); | ||
| w1[3] = hc_swap32 (w1[3]); | ||
| w2[0] = hc_swap32 (w2[0]); | ||
|
|
||
| /** | ||
| * prepend salt | ||
| */ | ||
|
|
||
| const u32x out_salt_len = out_len + salt_len; | ||
|
|
||
| u32x w0_t[4]; | ||
| u32x w1_t[4]; | ||
| u32x w2_t[4]; | ||
| u32x w3_t[4]; | ||
|
|
||
| w0_t[0] = salt_buf0[0]; | ||
| w0_t[1] = salt_buf0[1]; | ||
| w0_t[2] = w0[0]; | ||
| w0_t[3] = w0[1]; | ||
| w1_t[0] = w0[2]; | ||
| w1_t[1] = w0[3]; | ||
| w1_t[2] = w1[0]; | ||
| w1_t[3] = w1[1]; | ||
| w2_t[0] = w1[2]; | ||
| w2_t[1] = w1[3]; | ||
| w2_t[2] = w2[0]; | ||
| w2_t[3] = 0; | ||
| w3_t[0] = 0; | ||
| w3_t[1] = 0; | ||
| w3_t[2] = 0; | ||
| w3_t[3] = (out_salt_len + 1) * 8; | ||
|
|
||
| u64x digest[8]; | ||
|
|
||
| digest[0] = SHA512M_A; | ||
| digest[1] = SHA512M_B; | ||
| digest[2] = SHA512M_C; | ||
| digest[3] = SHA512M_D; | ||
| digest[4] = SHA512M_E; | ||
| digest[5] = SHA512M_F; | ||
| digest[6] = SHA512M_G; | ||
| digest[7] = SHA512M_H; | ||
|
|
||
| sha512_transform_intern (w0_t, w1_t, w2_t, w3_t, digest); | ||
|
|
||
| const u32x r0 = l32_from_64 (digest[7]); | ||
| const u32x r1 = h32_from_64 (digest[7]); | ||
| const u32x r2 = l32_from_64 (digest[3]); | ||
| const u32x r3 = h32_from_64 (digest[3]); | ||
|
|
||
| COMPARE_S_SIMD (r0, r1, r2, r3); | ||
| } | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_s08 (KERN_ATTR_RULES ()) | ||
| { | ||
| } | ||
|
|
||
| KERNEL_FQ void m22200_s16 (KERN_ATTR_RULES ()) | ||
| { | ||
| } |
Oops, something went wrong.
0 comments on commit
53105ab