From 9c6b533efd0612385aa6284ef304168988510a21 Mon Sep 17 00:00:00 2001 From: Zgzorx Date: Thu, 16 Feb 2023 11:03:10 +0100 Subject: [PATCH] Added sm3 a1 and a3 pure first draft --- OpenCL/m36000_a1-pure.cl | 111 +++++++++++++++++++++++++++++++ OpenCL/m36000_a3-pure.cl | 137 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 248 insertions(+) create mode 100644 OpenCL/m36000_a1-pure.cl create mode 100644 OpenCL/m36000_a3-pure.cl diff --git a/OpenCL/m36000_a1-pure.cl b/OpenCL/m36000_a1-pure.cl new file mode 100644 index 000000000..aa544bc80 --- /dev/null +++ b/OpenCL/m36000_a1-pure.cl @@ -0,0 +1,111 @@ +/** + * Author......: See docs/credits.txt + * License.....: MIT + */ + +//#define NEW_SIMD_CODE + +#ifdef KERNEL_STATIC +#include M2S(INCLUDE_PATH/inc_vendor.h) +#include M2S(INCLUDE_PATH/inc_types.h) +#include M2S(INCLUDE_PATH/inc_platform.cl) +#include M2S(INCLUDE_PATH/inc_common.cl) +#include M2S(INCLUDE_PATH/inc_scalar.cl) +#include M2S(INCLUDE_PATH/inc_hash_sm3.cl) +#endif + +KERNEL_FQ void m36000_mxx (KERN_ATTR_BASIC ()) +{ + /** + * modifier + */ + + const u64 lid = get_local_id (0); + const u64 gid = get_global_id (0); + + if (gid >= GID_CNT) return; + + /** + * base + */ + + sm3_ctx_t ctx0; + + sm3_init (&ctx0); + + sm3_update_global_swap (&ctx0, pws[gid].i, pws[gid].pw_len); + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++) + { + sm3_ctx_t ctx = ctx0; + + sm3_update_global_swap (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + sm3_final (&ctx); + + const u32 r0 = ctx.h[DGST_R0]; + const u32 r1 = ctx.h[DGST_R1]; + const u32 r2 = ctx.h[DGST_R2]; + const u32 r3 = ctx.h[DGST_R3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m36000_sxx (KERN_ATTR_BASIC ()) +{ + /** + * modifier + */ + + const u64 lid = get_local_id (0); + const u64 gid = get_global_id (0); + + if (gid >= GID_CNT) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R0], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R1], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R2], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R3] + }; + + /** + * base + */ + + sm3_ctx_t ctx0; + + sm3_init (&ctx0); + + sm3_update_global_swap (&ctx0, pws[gid].i, pws[gid].pw_len); + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++) + { + sm3_ctx_t ctx = ctx0; + + sm3_update_global_swap (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + sm3_final (&ctx); + + const u32 r0 = ctx.h[DGST_R0]; + const u32 r1 = ctx.h[DGST_R1]; + const u32 r2 = ctx.h[DGST_R2]; + const u32 r3 = ctx.h[DGST_R3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); + } +} diff --git a/OpenCL/m36000_a3-pure.cl b/OpenCL/m36000_a3-pure.cl new file mode 100644 index 000000000..9c205e6de --- /dev/null +++ b/OpenCL/m36000_a3-pure.cl @@ -0,0 +1,137 @@ +/** + * Author......: See docs/credits.txt + * License.....: MIT + */ + +#define NEW_SIMD_CODE + +#ifdef KERNEL_STATIC +#include M2S(INCLUDE_PATH/inc_vendor.h) +#include M2S(INCLUDE_PATH/inc_types.h) +#include M2S(INCLUDE_PATH/inc_platform.cl) +#include M2S(INCLUDE_PATH/inc_common.cl) +#include M2S(INCLUDE_PATH/inc_simd.cl) +#include M2S(INCLUDE_PATH/inc_hash_sm3.cl) +#endif + +KERNEL_FQ void m36000_mxx (KERN_ATTR_VECTOR ()) +{ + /** + * modifier + */ + + const u64 lid = get_local_id (0); + const u64 gid = get_global_id (0); + + if (gid >= GID_CNT) return; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < IL_CNT; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + sm3_ctx_vector_t ctx; + + sm3_init_vector (&ctx); + + sm3_update_vector (&ctx, w, pw_len); + + sm3_final_vector (&ctx); + + const u32x r0 = ctx.h[DGST_R0]; + const u32x r1 = ctx.h[DGST_R1]; + const u32x r2 = ctx.h[DGST_R2]; + const u32x r3 = ctx.h[DGST_R3]; + + COMPARE_M_SIMD (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m36000_sxx (KERN_ATTR_VECTOR ()) +{ + /** + * modifier + */ + + const u64 lid = get_local_id (0); + const u64 gid = get_global_id (0); + + if (gid >= GID_CNT) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R0], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R1], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R2], + digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R3] + }; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < IL_CNT; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + sm3_ctx_vector_t ctx; + + sm3_init_vector (&ctx); + + sm3_update_vector (&ctx, w, pw_len); + + sm3_final_vector (&ctx); + + const u32x r0 = ctx.h[DGST_R0]; + const u32x r1 = ctx.h[DGST_R1]; + const u32x r2 = ctx.h[DGST_R2]; + const u32x r3 = ctx.h[DGST_R3]; + + COMPARE_S_SIMD (r0, r1, r2, r3); + } +}