popov/cpuminer-opt-gpu
1
0
Fork 0
mirror of https://github.com/JayDDee/cpuminer-opt.git synced 2025-09-17 23:44:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

v3.19.1

Browse Source
This commit is contained in:
Jay D Dee
2021-11-20 00:46:01 -05:00
parent e6fd9b1d69
commit 7d2ef7973d
21 changed files with 186 additions and 1143 deletions
Download Patch File Download Diff File Expand all files Collapse all files

70
algo/swifftx/swifftx.c
View File

@@ -604,21 +604,14 @@ void InitializeSWIFFTX()
int omegaPowers[2 * N];
omegaPowers[0] = 1;
if (wasSetupDone)
return;
if (wasSetupDone) return;
for (i = 1; i < (2 * N); ++i)
{
omegaPowers[i] = Center(omegaPowers[i - 1] * OMEGA);
}
for (i = 0; i < (N / W); ++i)
{
for (j = 0; j < W; ++j)
{
multipliers[(i << 3) + j] = omegaPowers[ReverseBits(i, N / W) * (2 * j + 1)];
}
}
for (x = 0; x < 256; ++x)
{
@@ -626,10 +619,8 @@ void InitializeSWIFFTX()
{
register int temp = 0;
for (k = 0; k < 8; ++k)
{
temp += omegaPowers[(EIGHTH_N * (2 * j + 1) * ReverseBits(k, W)) % (2 * N)]
* ((x >> k) & 1);
}
fftTable[(x << 3) + j] = Center(temp);
}
@@ -703,18 +694,18 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
#if defined (__AVX512VL__) && defined(__AVX512BW__)
#define Q_REDUCE( a ) \
_mm256_sub_epi32( _mm256_and_si256( a, \
_mm256_movm_epi8( 0x11111111 ) ), _mm256_srai_epi32( a, 8 ) )
const __m256i mask = _mm256_movm_epi8( 0x11111111 );
#else
#else
#define Q_REDUCE( a ) \
_mm256_sub_epi32( _mm256_and_si256( a, \
m256_const1_32( 0x000000ff ) ), _mm256_srai_epi32( a, 8 ) )
const __m256i mask = m256_const1_32( 0x000000ff );
#endif
#define Q_REDUCE( a ) \
_mm256_sub_epi32( _mm256_and_si256( a, mask ), \
_mm256_srai_epi32( a, 8 ) )
out[0] = Q_REDUCE( F[0] );
out[1] = Q_REDUCE( F[1] );
out[2] = Q_REDUCE( F[2] );
@@ -805,9 +796,10 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
#undef ADD_SUB
const __m128i mask = m128_const1_32( 0x000000ff );
#define Q_REDUCE( a ) \
_mm_sub_epi32( _mm_and_si128( a, \
m128_const1_32( 0x000000ff ) ), _mm_srai_epi32( a, 8 ) )
_mm_sub_epi32( _mm_and_si128( a, mask ), _mm_srai_epi32( a, 8 ) )
out[ 0] = Q_REDUCE( F[ 0] );
out[ 1] = Q_REDUCE( F[ 1] );
@@ -1357,6 +1349,7 @@ void SWIFFTSum( const swift_int32_t *input, int m, unsigned char *output,
output[N] = carry;
}
/*
void ComputeSingleSWIFFTX_smooth(unsigned char input[SWIFFTX_INPUT_BLOCK_SIZE],
unsigned char output[SWIFFTX_OUTPUT_BLOCK_SIZE],
bool doSmooth)
@@ -1434,51 +1427,50 @@ void ComputeSingleSWIFFTX_smooth(unsigned char input[SWIFFTX_INPUT_BLOCK_SIZE],
output[N] = 0;
}
}
*/
void ComputeSingleSWIFFTX( unsigned char input[SWIFFTX_INPUT_BLOCK_SIZE],
unsigned char output[SWIFFTX_OUTPUT_BLOCK_SIZE] )
void ComputeSingleSWIFFTX( unsigned char *input, unsigned char *output )
{
int i;
// Will store the result of the FFT parts:
swift_int32_t fftOut[N * M] __attribute__ ((aligned (64)));
unsigned char intermediate[N * 3 + 8] __attribute__ ((aligned (64)));
unsigned char sum[ N*3 + 8 ] __attribute__ ((aligned (64)));
unsigned char carry0,carry1,carry2;
// Do the three SWIFFTS while remembering the three carry bytes (each carry byte gets
// overriden by the following SWIFFT):
// 1. Compute the FFT of the input - the common part for the first 3 SWIFFTs:
SWIFFTFFT(input, M, fftOut);
SWIFFTFFT( input, M, fftOut );
// 2. Compute the sums of the 3 SWIFFTs, each using a different set of coefficients:
// 2a. The first SWIFFT:
SWIFFTSum(fftOut, M, intermediate, As);
// Remember the carry byte:
carry0 = intermediate[N];
SWIFFTSum( fftOut, M, sum, As );
carry0 = sum[N];
// 2b. The second one:
SWIFFTSum(fftOut, M, intermediate + N, As + (M * N));
carry1 = intermediate[2 * N];
SWIFFTSum( fftOut, M, sum + N, As + M*N );
carry1 = sum[ 2*N ];
// 2c. The third one:
SWIFFTSum(fftOut, M, intermediate + (2 * N), As + 2 * (M * N));
carry2 = intermediate[3 * N];
SWIFFTSum( fftOut, M, sum + 2*N, As + 2*M*N );
carry2 = sum[ 3*N ];
//2d. Put three carry bytes in their place
intermediate[3 * N] = carry0;
intermediate[(3 * N) + 1] = carry1;
intermediate[(3 * N) + 2] = carry2;
sum[ 3*N ] = carry0;
sum[ 3*N + 1 ] = carry1;
sum[ 3*N + 2 ] = carry2;
// Padding intermediate output with 5 zeroes.
memset(intermediate + (3 * N) + 3, 0, 5);
memset( sum + 3*N + 3, 0, 5 );
// Apply the S-Box:
for ( i = 0; i < (3 * N) + 8; ++i )
intermediate[i] = SBox[intermediate[i]];
sum[i] = SBox[ sum[i] ];
// 3. The final and last SWIFFT:
SWIFFTFFT(intermediate, 3 * (N/8) + 1, fftOut);
SWIFFTSum(fftOut, 3 * (N/8) + 1, output, As);
SWIFFTFFT( sum, 3 * (N/8) + 1, fftOut );
SWIFFTSum( fftOut, 3 * (N/8) + 1, sum, As );
memcpy( output, sum, SWIFFTX_OUTPUT_BLOCK_SIZE - 1 );
}