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

v3.23.0

Browse Source
This commit is contained in:
Jay D Dee
2023-08-30 20:15:48 -04:00
parent 57a6b7b58b
commit 4378d2f841
72 changed files with 10184 additions and 2182 deletions
Download Patch File Download Diff File Expand all files Collapse all files

697
algo/swifftx/swifftx.c
View File

@@ -630,36 +630,35 @@ void InitializeSWIFFTX()
}
// In the original code the F matrix is rotated so it was not aranged
// the same as all the other data. Rearanging F to match all the other
// data made vectorizing possible, the compiler probably could have been
// able to auto-vectorize with proper data organisation.
// Also in the original code the custom 16 bit data types are all now 32
// bit int32_t regardless of the type name.
//
void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
// the same as the other data. Rearanging F made vectorizing up to 256 bits
// possible.
// Also in the original code the custom 16 bit data types are all now aliased
// to 32 bit int32_t.
void FFT( const unsigned char input[EIGHTH_N], swift_int32_t *output )
{
#if defined(__AVX2__)
__m256i F[8] __attribute__ ((aligned (64)));
__m256i F0, F1, F2, F3, F4, F5, F6, F7;
__m256i tbl = *(__m256i*)&( fftTable[ input[0] << 3 ] );
__m256i *mul = (__m256i*)multipliers;
__m256i *out = (__m256i*)output;
__m256i *tbl = (__m256i*)&( fftTable[ input[0] << 3 ] );
F[0] = _mm256_mullo_epi32( mul[0], *tbl );
tbl = (__m256i*)&( fftTable[ input[1] << 3 ] );
F[1] = _mm256_mullo_epi32( mul[1], *tbl );
tbl = (__m256i*)&( fftTable[ input[2] << 3 ] );
F[2] = _mm256_mullo_epi32( mul[2], *tbl );
tbl = (__m256i*)&( fftTable[ input[3] << 3 ] );
F[3] = _mm256_mullo_epi32( mul[3], *tbl );
tbl = (__m256i*)&( fftTable[ input[4] << 3 ] );
F[4] = _mm256_mullo_epi32( mul[4], *tbl );
tbl = (__m256i*)&( fftTable[ input[5] << 3 ] );
F[5] = _mm256_mullo_epi32( mul[5], *tbl );
tbl = (__m256i*)&( fftTable[ input[6] << 3 ] );
F[6] = _mm256_mullo_epi32( mul[6], *tbl );
tbl = (__m256i*)&( fftTable[ input[7] << 3 ] );
F[7] = _mm256_mullo_epi32( mul[7], *tbl );
F0 = _mm256_mullo_epi32( mul[0], tbl );
tbl = *(__m256i*)&( fftTable[ input[1] << 3 ] );
F1 = _mm256_mullo_epi32( mul[1], tbl );
tbl = *(__m256i*)&( fftTable[ input[2] << 3 ] );
F2 = _mm256_mullo_epi32( mul[2], tbl );
tbl = *(__m256i*)&( fftTable[ input[3] << 3 ] );
F3 = _mm256_mullo_epi32( mul[3], tbl );
tbl = *(__m256i*)&( fftTable[ input[4] << 3 ] );
F4 = _mm256_mullo_epi32( mul[4], tbl );
tbl = *(__m256i*)&( fftTable[ input[5] << 3 ] );
F5 = _mm256_mullo_epi32( mul[5], tbl );
tbl = *(__m256i*)&( fftTable[ input[6] << 3 ] );
F6 = _mm256_mullo_epi32( mul[6], tbl );
tbl = *(__m256i*)&( fftTable[ input[7] << 3 ] );
F7 = _mm256_mullo_epi32( mul[7], tbl );
#define ADD_SUB( a, b ) \
{ \
@@ -668,52 +667,50 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
a = _mm256_add_epi32( a, tmp ); \
}
ADD_SUB( F[0], F[1] );
ADD_SUB( F[2], F[3] );
ADD_SUB( F[4], F[5] );
ADD_SUB( F[6], F[7] );
F[3] = _mm256_slli_epi32( F[3], 4 );
F[7] = _mm256_slli_epi32( F[7], 4 );
ADD_SUB( F[0], F[2] );
ADD_SUB( F[1], F[3] );
ADD_SUB( F[4], F[6] );
ADD_SUB( F[5], F[7] );
F[5] = _mm256_slli_epi32( F[5], 2 );
F[6] = _mm256_slli_epi32( F[6], 4 );
F[7] = _mm256_slli_epi32( F[7], 6 );
ADD_SUB( F[0], F[4] );
ADD_SUB( F[1], F[5] );
ADD_SUB( F[2], F[6] );
ADD_SUB( F[3], F[7] );
ADD_SUB( F0, F1 );
ADD_SUB( F2, F3 );
ADD_SUB( F4, F5 );
ADD_SUB( F6, F7 );
F3 = _mm256_slli_epi32( F3, 4 );
F7 = _mm256_slli_epi32( F7, 4 );
ADD_SUB( F0, F2 );
ADD_SUB( F1, F3 );
ADD_SUB( F4, F6 );
ADD_SUB( F5, F7 );
F5 = _mm256_slli_epi32( F5, 2 );
F6 = _mm256_slli_epi32( F6, 4 );
F7 = _mm256_slli_epi32( F7, 6 );
ADD_SUB( F0, F4 );
ADD_SUB( F1, F5 );
ADD_SUB( F2, F6 );
ADD_SUB( F3, F7 );
#undef ADD_SUB
#if defined (__AVX512VL__) && defined(__AVX512BW__)
const __m256i mask = _mm256_movm_epi8( 0x11111111 );
#define Q_REDUCE( a ) \
_mm256_sub_epi32( _mm256_maskz_mov_epi8( 0x11111111, a ), \
_mm256_srai_epi32( a, 8 ) )
#else
const __m256i mask = m256_const1_32( 0x000000ff );
#endif
const __m256i mask = _mm256_set1_epi32( 0x000000ff );
#define Q_REDUCE( a ) \
_mm256_sub_epi32( _mm256_and_si256( a, mask ), \
_mm256_srai_epi32( a, 8 ) )
#endif
out[0] = Q_REDUCE( F[0] );
out[1] = Q_REDUCE( F[1] );
out[2] = Q_REDUCE( F[2] );
out[3] = Q_REDUCE( F[3] );
out[4] = Q_REDUCE( F[4] );
out[5] = Q_REDUCE( F[5] );
out[6] = Q_REDUCE( F[6] );
out[7] = Q_REDUCE( F[7] );
out[0] = Q_REDUCE( F0 );
out[1] = Q_REDUCE( F1 );
out[2] = Q_REDUCE( F2 );
out[3] = Q_REDUCE( F3 );
out[4] = Q_REDUCE( F4 );
out[5] = Q_REDUCE( F5 );
out[6] = Q_REDUCE( F6 );
out[7] = Q_REDUCE( F7 );
#undef Q_REDUCE
@@ -763,12 +760,10 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
ADD_SUB( F[ 9], F[11] );
ADD_SUB( F[12], F[14] );
ADD_SUB( F[13], F[15] );
F[ 6] = _mm_slli_epi32( F[ 6], 4 );
F[ 7] = _mm_slli_epi32( F[ 7], 4 );
F[14] = _mm_slli_epi32( F[14], 4 );
F[15] = _mm_slli_epi32( F[15], 4 );
ADD_SUB( F[ 0], F[ 4] );
ADD_SUB( F[ 1], F[ 5] );
ADD_SUB( F[ 2], F[ 6] );
@@ -777,14 +772,12 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
ADD_SUB( F[ 9], F[13] );
ADD_SUB( F[10], F[14] );
ADD_SUB( F[11], F[15] );
F[10] = _mm_slli_epi32( F[10], 2 );
F[11] = _mm_slli_epi32( F[11], 2 );
F[12] = _mm_slli_epi32( F[12], 4 );
F[13] = _mm_slli_epi32( F[13], 4 );
F[14] = _mm_slli_epi32( F[14], 6 );
F[15] = _mm_slli_epi32( F[15], 6 );
ADD_SUB( F[ 0], F[ 8] );
ADD_SUB( F[ 1], F[ 9] );
ADD_SUB( F[ 2], F[10] );
@@ -796,7 +789,7 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
#undef ADD_SUB
const __m128i mask = m128_const1_32( 0x000000ff );
const __m128i mask = _mm_set1_epi32( 0x000000ff );
#define Q_REDUCE( a ) \
_mm_sub_epi32( _mm_and_si128( a, mask ), _mm_srai_epi32( a, 8 ) )
@@ -820,16 +813,13 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
#undef Q_REDUCE
#else // < SSE4.1
#else // AVX256 elif SSE4_1
swift_int16_t *mult = multipliers;
// First loop unrolling:
register swift_int16_t *table = &(fftTable[input[0] << 3]);
/*
swift_int16_t *table = &( fftTable[ input[0] << 3 ] );
swift_int32_t F[64];
/*
for (int i = 0; i < 8; i++)
{
int j = i<<3;
@@ -845,99 +835,91 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
}
*/
register swift_int32_t F0, F1, F2, F3, F4, F5, F6, F7, F8, F9,
F10, F11, F12, F13, F14, F15, F16, F17, F18, F19,
F20, F21, F22, F23, F24, F25, F26, F27, F28, F29,
F30, F31, F32, F33, F34, F35, F36, F37, F38, F39,
F40, F41, F42, F43, F44, F45, F46, F47, F48, F49,
F50, F51, F52, F53, F54, F55, F56, F57, F58, F59,
F60, F61, F62, F63;
F0 = mult[0] * table[0];
F8 = mult[1] * table[1];
F16 = mult[2] * table[2];
F24 = mult[3] * table[3];
F32 = mult[4] * table[4];
F40 = mult[5] * table[5];
F48 = mult[6] * table[6];
F56 = mult[7] * table[7];
F[ 0] = mult[ 0] * table[0];
F[ 8] = mult[ 1] * table[1];
F[16] = mult[ 2] * table[2];
F[24] = mult[ 3] * table[3];
F[32] = mult[ 4] * table[4];
F[40] = mult[ 5] * table[5];
F[48] = mult[ 6] * table[6];
F[56] = mult[ 7] * table[7];
table = &(fftTable[input[1] << 3]);
F1 = mult[ 8] * table[0];
F9 = mult[ 9] * table[1];
F17 = mult[10] * table[2];
F25 = mult[11] * table[3];
F33 = mult[12] * table[4];
F41 = mult[13] * table[5];
F49 = mult[14] * table[6];
F57 = mult[15] * table[7];
F[ 1] = mult[ 8] * table[0];
F[ 9] = mult[ 9] * table[1];
F[17] = mult[10] * table[2];
F[25] = mult[11] * table[3];
F[33] = mult[12] * table[4];
F[41] = mult[13] * table[5];
F[49] = mult[14] * table[6];
F[57] = mult[15] * table[7];
table = &(fftTable[input[2] << 3]);
F2 = mult[16] * table[0];
F10 = mult[17] * table[1];
F18 = mult[18] * table[2];
F26 = mult[19] * table[3];
F34 = mult[20] * table[4];
F42 = mult[21] * table[5];
F50 = mult[22] * table[6];
F58 = mult[23] * table[7];
F[ 2] = mult[16] * table[0];
F[10] = mult[17] * table[1];
F[18] = mult[18] * table[2];
F[26] = mult[19] * table[3];
F[34] = mult[20] * table[4];
F[42] = mult[21] * table[5];
F[50] = mult[22] * table[6];
F[58] = mult[23] * table[7];
table = &(fftTable[input[3] << 3]);
F3 = mult[24] * table[0];
F11 = mult[25] * table[1];
F19 = mult[26] * table[2];
F27 = mult[27] * table[3];
F35 = mult[28] * table[4];
F43 = mult[29] * table[5];
F51 = mult[30] * table[6];
F59 = mult[31] * table[7];
F[ 3] = mult[24] * table[0];
F[11] = mult[25] * table[1];
F[19] = mult[26] * table[2];
F[27] = mult[27] * table[3];
F[35] = mult[28] * table[4];
F[43] = mult[29] * table[5];
F[51] = mult[30] * table[6];
F[59] = mult[31] * table[7];
table = &(fftTable[input[4] << 3]);
F4 = mult[32] * table[0];
F12 = mult[33] * table[1];
F20 = mult[34] * table[2];
F28 = mult[35] * table[3];
F36 = mult[36] * table[4];
F44 = mult[37] * table[5];
F52 = mult[38] * table[6];
F60 = mult[39] * table[7];
F[ 4] = mult[32] * table[0];
F[12] = mult[33] * table[1];
F[20] = mult[34] * table[2];
F[28] = mult[35] * table[3];
F[36] = mult[36] * table[4];
F[44] = mult[37] * table[5];
F[52] = mult[38] * table[6];
F[60] = mult[39] * table[7];
table = &(fftTable[input[5] << 3]);
F5 = mult[40] * table[0];
F13 = mult[41] * table[1];
F21 = mult[42] * table[2];
F29 = mult[43] * table[3];
F37 = mult[44] * table[4];
F45 = mult[45] * table[5];
F53 = mult[46] * table[6];
F61 = mult[47] * table[7];
F[ 5] = mult[40] * table[0];
F[13] = mult[41] * table[1];
F[21] = mult[42] * table[2];
F[29] = mult[43] * table[3];
F[37] = mult[44] * table[4];
F[45] = mult[45] * table[5];
F[53] = mult[46] * table[6];
F[61] = mult[47] * table[7];
table = &(fftTable[input[6] << 3]);
F6 = mult[48] * table[0];
F14 = mult[49] * table[1];
F22 = mult[50] * table[2];
F30 = mult[51] * table[3];
F38 = mult[52] * table[4];
F46 = mult[53] * table[5];
F54 = mult[54] * table[6];
F62 = mult[55] * table[7];
F[ 6] = mult[48] * table[0];
F[14] = mult[49] * table[1];
F[22] = mult[50] * table[2];
F[30] = mult[51] * table[3];
F[38] = mult[52] * table[4];
F[46] = mult[53] * table[5];
F[54] = mult[54] * table[6];
F[62] = mult[55] * table[7];
table = &(fftTable[input[7] << 3]);
F7 = mult[56] * table[0];
F15 = mult[57] * table[1];
F23 = mult[58] * table[2];
F31 = mult[59] * table[3];
F39 = mult[60] * table[4];
F47 = mult[61] * table[5];
F55 = mult[62] * table[6];
F63 = mult[63] * table[7];
F[ 7] = mult[56] * table[0];
F[15] = mult[57] * table[1];
F[23] = mult[58] * table[2];
F[31] = mult[59] * table[3];
F[39] = mult[60] * table[4];
F[47] = mult[61] * table[5];
F[55] = mult[62] * table[6];
F[63] = mult[63] * table[7];
#define ADD_SUB( a, b ) \
{ \
@@ -987,262 +969,229 @@ void FFT(const unsigned char input[EIGHTH_N], swift_int32_t *output)
}
*/
// Second loop unrolling:
// Iteration 0:
ADD_SUB(F0, F1);
ADD_SUB(F2, F3);
ADD_SUB(F4, F5);
ADD_SUB(F6, F7);
ADD_SUB( F[ 0], F[ 1] );
ADD_SUB( F[ 2], F[ 3] );
ADD_SUB( F[ 4], F[ 5] );
ADD_SUB( F[ 6], F[ 7] );
F[ 3] <<= 4;
F[ 7] <<= 4;
ADD_SUB( F[ 0], F[ 2] );
ADD_SUB( F[ 1], F[ 3] );
ADD_SUB( F[ 4], F[ 6] );
ADD_SUB( F[ 5], F[ 7] );
F[ 5] <<= 2;
F[ 6] <<= 4;
F[ 7] <<= 6;
ADD_SUB( F[ 0], F[ 4] );
ADD_SUB( F[ 1], F[ 5] );
ADD_SUB( F[ 2], F[ 6] );
ADD_SUB( F[ 3], F[ 7] );
F3 <<= 4;
F7 <<= 4;
ADD_SUB(F0, F2);
ADD_SUB(F1, F3);
ADD_SUB(F4, F6);
ADD_SUB(F5, F7);
F5 <<= 2;
F6 <<= 4;
F7 <<= 6;
ADD_SUB(F0, F4);
ADD_SUB(F1, F5);
ADD_SUB(F2, F6);
ADD_SUB(F3, F7);
output[0] = Q_REDUCE(F0);
output[8] = Q_REDUCE(F1);
output[16] = Q_REDUCE(F2);
output[24] = Q_REDUCE(F3);
output[32] = Q_REDUCE(F4);
output[40] = Q_REDUCE(F5);
output[48] = Q_REDUCE(F6);
output[56] = Q_REDUCE(F7);
output[ 0] = Q_REDUCE( F[ 0] );
output[ 8] = Q_REDUCE( F[ 1] );
output[16] = Q_REDUCE( F[ 2] );
output[24] = Q_REDUCE( F[ 3] );
output[32] = Q_REDUCE( F[ 4] );
output[40] = Q_REDUCE( F[ 5] );
output[48] = Q_REDUCE( F[ 6] );
output[56] = Q_REDUCE( F[ 7] );
// Iteration 1:
ADD_SUB(F8, F9);
ADD_SUB(F10, F11);
ADD_SUB(F12, F13);
ADD_SUB(F14, F15);
ADD_SUB( F[ 8], F[ 9] );
ADD_SUB( F[10], F[11] );
ADD_SUB( F[12], F[13] );
ADD_SUB( F[14], F[15] );
F[11] <<= 4;
F[15] <<= 4;
ADD_SUB( F[ 8], F[10] );
ADD_SUB( F[ 9], F[11] );
ADD_SUB( F[12], F[14] );
ADD_SUB( F[13], F[15] );
F[13] <<= 2;
F[14] <<= 4;
F[15] <<= 6;
ADD_SUB( F[ 8], F[12] );
ADD_SUB( F[ 9], F[13] );
ADD_SUB( F[10], F[14] );
ADD_SUB( F[11], F[15] );
F11 <<= 4;
F15 <<= 4;
ADD_SUB(F8, F10);
ADD_SUB(F9, F11);
ADD_SUB(F12, F14);
ADD_SUB(F13, F15);
F13 <<= 2;
F14 <<= 4;
F15 <<= 6;
ADD_SUB(F8, F12);
ADD_SUB(F9, F13);
ADD_SUB(F10, F14);
ADD_SUB(F11, F15);
output[1] = Q_REDUCE(F8);
output[9] = Q_REDUCE(F9);
output[17] = Q_REDUCE(F10);
output[25] = Q_REDUCE(F11);
output[33] = Q_REDUCE(F12);
output[41] = Q_REDUCE(F13);
output[49] = Q_REDUCE(F14);
output[57] = Q_REDUCE(F15);
output[ 1] = Q_REDUCE( F[ 8] );
output[ 9] = Q_REDUCE( F[ 9] );
output[17] = Q_REDUCE( F[10] );
output[25] = Q_REDUCE( F[11] );
output[33] = Q_REDUCE( F[12] );
output[41] = Q_REDUCE( F[13] );
output[49] = Q_REDUCE( F[14] );
output[57] = Q_REDUCE( F[15] );
// Iteration 2:
ADD_SUB(F16, F17);
ADD_SUB(F18, F19);
ADD_SUB(F20, F21);
ADD_SUB(F22, F23);
ADD_SUB( F[16], F[17] );
ADD_SUB( F[18], F[19] );
ADD_SUB( F[20], F[21] );
ADD_SUB( F[22], F[23] );
F[19] <<= 4;
F[23] <<= 4;
ADD_SUB( F[16], F[18]);
ADD_SUB( F[17], F[19]);
ADD_SUB( F[20], F[22]);
ADD_SUB( F[21], F[23]);
F[21] <<= 2;
F[22] <<= 4;
F[23] <<= 6;
ADD_SUB( F[16], F[20] );
ADD_SUB( F[17], F[21] );
ADD_SUB( F[18], F[22] );
ADD_SUB( F[19], F[23] );
F19 <<= 4;
F23 <<= 4;
ADD_SUB(F16, F18);
ADD_SUB(F17, F19);
ADD_SUB(F20, F22);
ADD_SUB(F21, F23);
F21 <<= 2;
F22 <<= 4;
F23 <<= 6;
ADD_SUB(F16, F20);
ADD_SUB(F17, F21);
ADD_SUB(F18, F22);
ADD_SUB(F19, F23);
output[2] = Q_REDUCE(F16);
output[10] = Q_REDUCE(F17);
output[18] = Q_REDUCE(F18);
output[26] = Q_REDUCE(F19);
output[34] = Q_REDUCE(F20);
output[42] = Q_REDUCE(F21);
output[50] = Q_REDUCE(F22);
output[58] = Q_REDUCE(F23);
output[ 2] = Q_REDUCE( F[16] );
output[10] = Q_REDUCE( F[17] );
output[18] = Q_REDUCE( F[18] );
output[26] = Q_REDUCE( F[19] );
output[34] = Q_REDUCE( F[20] );
output[42] = Q_REDUCE( F[21] );
output[50] = Q_REDUCE( F[22] );
output[58] = Q_REDUCE( F[23] );
// Iteration 3:
ADD_SUB(F24, F25);
ADD_SUB(F26, F27);
ADD_SUB(F28, F29);
ADD_SUB(F30, F31);
ADD_SUB( F[24], F[25] );
ADD_SUB( F[26], F[27] );
ADD_SUB( F[28], F[29] );
ADD_SUB( F[30], F[31] );
F[27] <<= 4;
F[31] <<= 4;
ADD_SUB( F[24], F[26] );
ADD_SUB( F[25], F[27] );
ADD_SUB( F[28], F[30] );
ADD_SUB( F[29], F[31] );
F[29] <<= 2;
F[30] <<= 4;
F[31] <<= 6;
ADD_SUB( F[24], F[28] );
ADD_SUB( F[25], F[29] );
ADD_SUB( F[26], F[30] );
ADD_SUB( F[27], F[31] );
F27 <<= 4;
F31 <<= 4;
ADD_SUB(F24, F26);
ADD_SUB(F25, F27);
ADD_SUB(F28, F30);
ADD_SUB(F29, F31);
F29 <<= 2;
F30 <<= 4;
F31 <<= 6;
ADD_SUB(F24, F28);
ADD_SUB(F25, F29);
ADD_SUB(F26, F30);
ADD_SUB(F27, F31);
output[3] = Q_REDUCE(F24);
output[11] = Q_REDUCE(F25);
output[19] = Q_REDUCE(F26);
output[27] = Q_REDUCE(F27);
output[35] = Q_REDUCE(F28);
output[43] = Q_REDUCE(F29);
output[51] = Q_REDUCE(F30);
output[59] = Q_REDUCE(F31);
output[ 3] = Q_REDUCE( F[24] );
output[11] = Q_REDUCE( F[25] );
output[19] = Q_REDUCE( F[26] );
output[27] = Q_REDUCE( F[27] );
output[35] = Q_REDUCE( F[28] );
output[43] = Q_REDUCE( F[29] );
output[51] = Q_REDUCE( F[30] );
output[59] = Q_REDUCE( F[31] );
// Iteration 4:
ADD_SUB(F32, F33);
ADD_SUB(F34, F35);
ADD_SUB(F36, F37);
ADD_SUB(F38, F39);
ADD_SUB( F[32], F[33] );
ADD_SUB( F[34], F[35] );
ADD_SUB( F[36], F[37] );
ADD_SUB( F[38], F[39] );
F[35] <<= 4;
F[39] <<= 4;
ADD_SUB( F[32], F[34] );
ADD_SUB( F[33], F[35] );
ADD_SUB( F[36], F[38] );
ADD_SUB( F[37], F[39] );
F[37] <<= 2;
F[38] <<= 4;
F[39] <<= 6;
ADD_SUB( F[32], F[36] );
ADD_SUB( F[33], F[37] );
ADD_SUB( F[34], F[38] );
ADD_SUB( F[35], F[39] );
F35 <<= 4;
F39 <<= 4;
ADD_SUB(F32, F34);
ADD_SUB(F33, F35);
ADD_SUB(F36, F38);
ADD_SUB(F37, F39);
F37 <<= 2;
F38 <<= 4;
F39 <<= 6;
ADD_SUB(F32, F36);
ADD_SUB(F33, F37);
ADD_SUB(F34, F38);
ADD_SUB(F35, F39);
output[4] = Q_REDUCE(F32);
output[12] = Q_REDUCE(F33);
output[20] = Q_REDUCE(F34);
output[28] = Q_REDUCE(F35);
output[36] = Q_REDUCE(F36);
output[44] = Q_REDUCE(F37);
output[52] = Q_REDUCE(F38);
output[60] = Q_REDUCE(F39);
output[ 4] = Q_REDUCE( F[32] );
output[12] = Q_REDUCE( F[33] );
output[20] = Q_REDUCE( F[34] );
output[28] = Q_REDUCE( F[35] );
output[36] = Q_REDUCE( F[36] );
output[44] = Q_REDUCE( F[37] );
output[52] = Q_REDUCE( F[38] );
output[60] = Q_REDUCE( F[39] );
// Iteration 5:
ADD_SUB(F40, F41);
ADD_SUB(F42, F43);
ADD_SUB(F44, F45);
ADD_SUB(F46, F47);
ADD_SUB( F[40], F[41] );
ADD_SUB( F[42], F[43] );
ADD_SUB( F[44], F[45] );
ADD_SUB( F[46], F[47] );
F[43] <<= 4;
F[47] <<= 4;
ADD_SUB( F[40], F[42] );
ADD_SUB( F[41], F[43] );
ADD_SUB( F[44], F[46] );
ADD_SUB( F[45], F[47] );
F[45] <<= 2;
F[46] <<= 4;
F[47] <<= 6;
ADD_SUB( F[40], F[44] );
ADD_SUB( F[41], F[45] );
ADD_SUB( F[42], F[46] );
ADD_SUB( F[43], F[47] );
F43 <<= 4;
F47 <<= 4;
ADD_SUB(F40, F42);
ADD_SUB(F41, F43);
ADD_SUB(F44, F46);
ADD_SUB(F45, F47);
F45 <<= 2;
F46 <<= 4;
F47 <<= 6;
ADD_SUB(F40, F44);
ADD_SUB(F41, F45);
ADD_SUB(F42, F46);
ADD_SUB(F43, F47);
output[5] = Q_REDUCE(F40);
output[13] = Q_REDUCE(F41);
output[21] = Q_REDUCE(F42);
output[29] = Q_REDUCE(F43);
output[37] = Q_REDUCE(F44);
output[45] = Q_REDUCE(F45);
output[53] = Q_REDUCE(F46);
output[61] = Q_REDUCE(F47);
output[ 5] = Q_REDUCE( F[40] );
output[13] = Q_REDUCE( F[41] );
output[21] = Q_REDUCE( F[42] );
output[29] = Q_REDUCE( F[43] );
output[37] = Q_REDUCE( F[44] );
output[45] = Q_REDUCE( F[45] );
output[53] = Q_REDUCE( F[46] );
output[61] = Q_REDUCE( F[47] );
// Iteration 6:
ADD_SUB(F48, F49);
ADD_SUB(F50, F51);
ADD_SUB(F52, F53);
ADD_SUB(F54, F55);
ADD_SUB( F[48], F[49] );
ADD_SUB( F[50], F[51] );
ADD_SUB( F[52], F[53] );
ADD_SUB( F[54], F[55] );
F[51] <<= 4;
F[55] <<= 4;
ADD_SUB( F[48], F[50] );
ADD_SUB( F[49], F[51] );
ADD_SUB( F[52], F[54] );
ADD_SUB( F[53], F[55] );
F[53] <<= 2;
F[54] <<= 4;
F[55] <<= 6;
ADD_SUB( F[48], F[52] );
ADD_SUB( F[49], F[53] );
ADD_SUB( F[50], F[54] );
ADD_SUB( F[51], F[55] );
F51 <<= 4;
F55 <<= 4;
ADD_SUB(F48, F50);
ADD_SUB(F49, F51);
ADD_SUB(F52, F54);
ADD_SUB(F53, F55);
F53 <<= 2;
F54 <<= 4;
F55 <<= 6;
ADD_SUB(F48, F52);
ADD_SUB(F49, F53);
ADD_SUB(F50, F54);
ADD_SUB(F51, F55);
output[6] = Q_REDUCE(F48);
output[14] = Q_REDUCE(F49);
output[22] = Q_REDUCE(F50);
output[30] = Q_REDUCE(F51);
output[38] = Q_REDUCE(F52);
output[46] = Q_REDUCE(F53);
output[54] = Q_REDUCE(F54);
output[62] = Q_REDUCE(F55);
output[ 6] = Q_REDUCE( F[48] );
output[14] = Q_REDUCE( F[49] );
output[22] = Q_REDUCE( F[50] );
output[30] = Q_REDUCE( F[51] );
output[38] = Q_REDUCE( F[52] );
output[46] = Q_REDUCE( F[53] );
output[54] = Q_REDUCE( F[54] );
output[62] = Q_REDUCE( F[55] );
// Iteration 7:
ADD_SUB(F56, F57);
ADD_SUB(F58, F59);
ADD_SUB(F60, F61);
ADD_SUB(F62, F63);
ADD_SUB( F[56], F[57] );
ADD_SUB( F[58], F[59] );
ADD_SUB( F[60], F[61] );
ADD_SUB( F[62], F[63] );
F[59] <<= 4;
F[63] <<= 4;
ADD_SUB( F[56], F[58] );
ADD_SUB( F[57], F[59] );
ADD_SUB( F[60], F[62] );
ADD_SUB( F[61], F[63] );
F[61] <<= 2;
F[62] <<= 4;
F[63] <<= 6;
ADD_SUB( F[56], F[60] );
ADD_SUB( F[57], F[61] );
ADD_SUB( F[58], F[62] );
ADD_SUB( F[59], F[63] );
F59 <<= 4;
F63 <<= 4;
ADD_SUB(F56, F58);
ADD_SUB(F57, F59);
ADD_SUB(F60, F62);
ADD_SUB(F61, F63);
F61 <<= 2;
F62 <<= 4;
F63 <<= 6;
ADD_SUB(F56, F60);
ADD_SUB(F57, F61);
ADD_SUB(F58, F62);
ADD_SUB(F59, F63);
output[7] = Q_REDUCE(F56);
output[15] = Q_REDUCE(F57);
output[23] = Q_REDUCE(F58);
output[31] = Q_REDUCE(F59);
output[39] = Q_REDUCE(F60);
output[47] = Q_REDUCE(F61);
output[55] = Q_REDUCE(F62);
output[63] = Q_REDUCE(F63);
output[ 7] = Q_REDUCE( F[56] );
output[15] = Q_REDUCE( F[57] );
output[23] = Q_REDUCE( F[58] );
output[31] = Q_REDUCE( F[59] );
output[39] = Q_REDUCE( F[60] );
output[47] = Q_REDUCE( F[61] );
output[55] = Q_REDUCE( F[62] );
output[63] = Q_REDUCE( F[63] );
#undef ADD_SUB
#undef Q_REDUCE