diff options
author | katsu560 <118887472+katsu560@users.noreply.github.com> | 2023-05-14 19:03:51 +0900 |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-05-14 10:03:51 +0000 |
commit | 60f8c361ca26328ef8523dfb08077fe2f1034490 (patch) | |
tree | 0693d67c6be687f92a17bfd0b5ee5b7242b5a654 | |
parent | 601a033475645370483973817d987928ea95f36c (diff) |
ggml : add AVX support based on AVX2 code (#1430)
-rw-r--r-- | ggml.c | 135 |
1 files changed, 132 insertions, 3 deletions
@@ -580,7 +580,63 @@ static inline __m128i packNibbles( __m256i bytes ) return _mm_packus_epi16( r0, r1 ); #endif } -#else +#elif defined(__AVX__) +// spread 32 bits to 32 bytes { 0x00, 0xFF } +static inline __m256i bytes_from_bits_32(const uint8_t * x) { + uint32_t x32; + memcpy(&x32, x, sizeof(uint32_t)); + const __m128i shuf_maskl = _mm_set_epi64x(0x0101010101010101, 0x0000000000000000); + const __m128i shuf_maskh = _mm_set_epi64x(0x0303030303030303, 0x0202020202020202); + __m128i bytesl = _mm_shuffle_epi8(_mm_set1_epi32(x32), shuf_maskl); + __m128i bytesh = _mm_shuffle_epi8(_mm_set1_epi32(x32), shuf_maskh); + const __m128i bit_mask = _mm_set1_epi64x(0x7fbfdfeff7fbfdfe); + bytesl = _mm_or_si128(bytesl, bit_mask); + bytesh = _mm_or_si128(bytesh, bit_mask); + bytesl = _mm_cmpeq_epi8(bytesl, _mm_set1_epi64x(-1)); + bytesh = _mm_cmpeq_epi8(bytesh, _mm_set1_epi64x(-1)); + return _mm256_set_m128i(bytesh, bytesl); +} + +// Unpack 32 4-bit fields into 32 bytes +// The output vector contains 32 bytes, each one in [ 0 .. 15 ] interval +static inline __m256i bytes_from_nibbles_32(const uint8_t * rsi) +{ + // Load 16 bytes from memory + __m128i tmpl = _mm_loadu_si128((const __m128i *)rsi); + __m128i tmph = _mm_srli_epi16(tmpl, 4); + const __m128i lowMask = _mm_set1_epi8(0xF); + tmpl = _mm_and_si128(lowMask, tmpl); + tmph = _mm_and_si128(lowMask, tmph); + return _mm256_set_m128i(tmph, tmpl); +} + +// add int16_t pairwise and return as float vector +static inline __m256 sum_i16_pairs_float(const __m128i xh, const __m128i xl) { + const __m128i ones = _mm_set1_epi16(1); + const __m128i summed_pairsl = _mm_madd_epi16(ones, xl); + const __m128i summed_pairsh = _mm_madd_epi16(ones, xh); + const __m256i summed_pairs = _mm256_set_m128i(summed_pairsh, summed_pairsl); + return _mm256_cvtepi32_ps(summed_pairs); +} + +// multiply int8_t, add results pairwise twice and return as float vector +static inline __m256 mul_sum_i8_pairs_float(const __m256i x, const __m256i y) { + const __m128i xl = _mm256_castsi256_si128(x); + const __m128i xh = _mm256_extractf128_si256(x, 1); + const __m128i yl = _mm256_castsi256_si128(y); + const __m128i yh = _mm256_extractf128_si256(y, 1); + // Get absolute values of x vectors + const __m128i axl = _mm_sign_epi8(xl, xl); + const __m128i axh = _mm_sign_epi8(xh, xh); + // Sign the values of the y vectors + const __m128i syl = _mm_sign_epi8(yl, xl); + const __m128i syh = _mm_sign_epi8(yh, xh); + // Perform multiplication and create 16-bit values + const __m128i dotl = _mm_maddubs_epi16(axl, syl); + const __m128i doth = _mm_maddubs_epi16(axh, syh); + return sum_i16_pairs_float(doth, dotl); +} + static inline __m128i packNibbles( __m128i bytes1, __m128i bytes2 ) { // Move bits within 16-bit lanes from 0000_abcd_0000_efgh into 0000_0000_abcd_efgh @@ -2355,7 +2411,7 @@ static void ggml_vec_dot_q4_1_q8_1(const int n, float * restrict s, const void * } *s = vaddvq_f32(sumv0) + vaddvq_f32(sumv1) + summs; -#elif defined(__AVX2__) +#elif defined(__AVX2__) || defined(__AVX__) // Initialize accumulator with zeros __m256 acc = _mm256_setzero_ps(); @@ -2381,7 +2437,11 @@ static void ggml_vec_dot_q4_1_q8_1(const int n, float * restrict s, const void * const __m256 xy = mul_sum_i8_pairs_float(bx, by); // Accumulate d0*d1*x*y +#if defined(__AVX2__) acc = _mm256_fmadd_ps( d0d1, xy, acc ); +#else + acc = _mm256_add_ps( _mm256_mul_ps( d0d1, xy ), acc ); +#endif } *s = hsum_float_8(acc) + summs; @@ -2593,6 +2653,37 @@ static void ggml_vec_dot_q5_0_q8_0(const int n, float * restrict s, const void * } *s = hsum_float_8(acc); +#elif defined(__AVX__) + // Initialize accumulator with zeros + __m256 acc = _mm256_setzero_ps(); + __m128i mask = _mm_set1_epi8((char)0xF0); + + // Main loop + for (int i = 0; i < nb; i++) { + /* Compute combined scale for the block */ + const __m256 d = _mm256_mul_ps(_mm256_set1_ps(GGML_FP16_TO_FP32(x[i].d)), _mm256_broadcast_ss(&y[i].d)); + + __m256i bx = bytes_from_nibbles_32(x[i].qs); + const __m256i bxhi = bytes_from_bits_32(x[i].qh); + __m128i bxhil = _mm256_castsi256_si128(bxhi); + __m128i bxhih = _mm256_extractf128_si256(bxhi, 1); + bxhil = _mm_andnot_si128(bxhil, mask); + bxhih = _mm_andnot_si128(bxhih, mask); + __m128i bxl = _mm256_castsi256_si128(bx); + __m128i bxh = _mm256_extractf128_si256(bx, 1); + bxl = _mm_or_si128(bxl, bxhil); + bxh = _mm_or_si128(bxh, bxhih); + bx = _mm256_set_m128i(bxh, bxl); + + const __m256i by = _mm256_loadu_si256((const __m256i *)y[i].qs); + + const __m256 q = mul_sum_i8_pairs_float(bx, by); + + /* Multiply q with scale and accumulate */ + acc = _mm256_add_ps(_mm256_mul_ps(d, q), acc); + } + + *s = hsum_float_8(acc); #else // scalar float sumf = 0.0; @@ -2821,6 +2912,40 @@ static void ggml_vec_dot_q5_1_q8_1(const int n, float * restrict s, const void * } *s = hsum_float_8(acc) + summs; +#elif defined(__AVX__) + // Initialize accumulator with zeros + __m256 acc = _mm256_setzero_ps(); + __m128i mask = _mm_set1_epi8(0x10); + + float summs = 0.0f; + + // Main loop + for (int i = 0; i < nb; i++) { + const __m256 dx = _mm256_set1_ps(GGML_FP16_TO_FP32(x[i].d)); + + summs += GGML_FP16_TO_FP32(x[i].m) * y[i].s; + + __m256i bx = bytes_from_nibbles_32(x[i].qs); + const __m256i bxhi = bytes_from_bits_32(x[i].qh); + __m128i bxhil = _mm256_castsi256_si128(bxhi); + __m128i bxhih = _mm256_extractf128_si256(bxhi, 1); + bxhil = _mm_and_si128(bxhil, mask); + bxhih = _mm_and_si128(bxhih, mask); + __m128i bxl = _mm256_castsi256_si128(bx); + __m128i bxh = _mm256_extractf128_si256(bx, 1); + bxl = _mm_or_si128(bxl, bxhil); + bxh = _mm_or_si128(bxh, bxhih); + bx = _mm256_set_m128i(bxh, bxl); + + const __m256 dy = _mm256_broadcast_ss(&y[i].d); + const __m256i by = _mm256_loadu_si256((const __m256i *)y[i].qs); + + const __m256 q = mul_sum_i8_pairs_float(bx, by); + + acc = _mm256_add_ps(_mm256_mul_ps(q, _mm256_mul_ps(dx, dy)), acc); + } + + *s = hsum_float_8(acc) + summs; #else // scalar float sumf = 0.0; @@ -2910,7 +3035,7 @@ static void ggml_vec_dot_q8_0_q8_0(const int n, float * restrict s, const void * } *s = vaddvq_f32(sumv0) + vaddvq_f32(sumv1); -#elif defined(__AVX2__) +#elif defined(__AVX2__) || defined(__AVX__) // Initialize accumulator with zeros __m256 acc = _mm256_setzero_ps(); @@ -2924,7 +3049,11 @@ static void ggml_vec_dot_q8_0_q8_0(const int n, float * restrict s, const void * const __m256 q = mul_sum_i8_pairs_float(bx, by); // Multiply q with scale and accumulate +#if defined(__AVX2__) acc = _mm256_fmadd_ps( d, q, acc ); +#else + acc = _mm256_add_ps( _mm256_mul_ps( d, q ), acc ); +#endif } *s = hsum_float_8(acc); |