GOSTSec/ccminer
1
0
Fork 0
mirror of https://github.com/GOSTSec/ccminer synced 2025-08-26 22:02:08 +00:00
Code Issues Releases Wiki Activity

lbry: small sha512 improvement

Browse Source
This commit is contained in:
Tanguy Pruvot 2016-09-04 18:06:10 +02:00
parent f8aa16f8d2
commit f982fd70d5
1 changed files with 20 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
lbry/cuda_sha512_lbry.cu
View File

@ -40,12 +40,21 @@ uint64_t K_512[80] = {
#undef xor3 #undef xor3
#define xor3(a,b,c) (a^b^c) #define xor3(a,b,c) (a^b^c)
//#define ROR64_8(x) ROTR64(x,8)
__device__ __inline__
uint64_t ROR64_8(const uint64_t u64) {
const uint2 a = vectorize(u64);
uint2 result;
result.x = __byte_perm(a.y, a.x, 0x0765);
result.y = __byte_perm(a.y, a.x, 0x4321);
return devectorize(result);
}
#define bsg5_0(x) xor3(ROTR64(x,28),ROTR64(x,34),ROTR64(x,39)) #define bsg5_0(x) xor3(ROTR64(x,28),ROTR64(x,34),ROTR64(x,39))
#define bsg5_1(x) xor3(ROTR64(x,14),ROTR64(x,18),ROTR64(x,41)) #define bsg5_1(x) xor3(ROTR64(x,14),ROTR64(x,18),ROTR64(x,41))
#define ssg5_0(x) xor3(ROTR64(x,1),ROTR64(x,8),x>>7) #define ssg5_0(x) xor3(ROTR64(x,1), ROR64_8(x), x>>7)
#define ssg5_1(x) xor3(ROTR64(x,19),ROTR64(x,61), x>>6) #define ssg5_1(x) xor3(ROTR64(x,19),ROTR64(x,61), x>>6)
#define andor64(a,b,c) ((a & (b | c)) | (b & c)) #define andor64(a,b,c) ((a & (b | c)) | (b & c))
#define xandx64(e,f,g) (g ^ (e & (g ^ f))) #define xandx64(e,f,g) (g ^ (e & (g ^ f)))
@ -67,6 +76,7 @@ void lbry_sha512_gpu_hash_32(const uint32_t threads, uint64_t *g_hash)
0x6A09E667F3BCC908, 0xBB67AE8584CAA73B, 0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1, 0x6A09E667F3BCC908, 0xBB67AE8584CAA73B, 0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1,
0x510E527FADE682D1, 0x9B05688C2B3E6C1F, 0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179 0x510E527FADE682D1, 0x9B05688C2B3E6C1F, 0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179
}; };
uint64_t r[8]; uint64_t r[8];
uint64_t W[16]; uint64_t W[16];
if (thread < threads) if (thread < threads)
@ -76,7 +86,7 @@ void lbry_sha512_gpu_hash_32(const uint32_t threads, uint64_t *g_hash)
*(uint2x4*)&r[0] = *(uint2x4*)&IV512[0]; *(uint2x4*)&r[0] = *(uint2x4*)&IV512[0];
*(uint2x4*)&r[4] = *(uint2x4*)&IV512[4]; *(uint2x4*)&r[4] = *(uint2x4*)&IV512[4];
*(uint2x4*)&W[ 0] = __ldg4((uint2x4*)&pHash[ 0]); *(uint2x4*)&W[0] = __ldg4((uint2x4*)pHash);
W[4] = 0x8000000000000000; // end tag W[4] = 0x8000000000000000; // end tag