Q4_1 quantization (#193)

* Add AVX2 version of ggml_vec_dot_q4_1

* Small optimisations to q4_1 dot product (@Const-me)

* Rearrange Q4_1 quantization to work for multipart models. (Fix #152)

* Fix ggml_vec_mad_q4_1 too

* Fix non-vectorised q4_1 vec mul
This commit is contained in:
Matvey Soloviev 2023-03-17 05:48:39 +01:00 committed by GitHub
parent 721311070e
commit 904d2a8d6a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 130 additions and 39 deletions

149
ggml.c
View file

@ -607,10 +607,11 @@ void quantize_row_q4_1(const float * restrict x, void * restrict y, int k) {
assert(k % QK == 0); assert(k % QK == 0);
const int nb = k / QK; const int nb = k / QK;
const size_t bs = 2*sizeof(float) + QK/2;
float * restrict pm = (float *) (y); uint8_t * restrict pd = ((uint8_t *)y + 0*bs);
float * restrict pd = (float *) (pm + nb); uint8_t * restrict pm = ((uint8_t *)y + 0*bs + sizeof(float));
uint8_t * restrict pb = (uint8_t *) (pd + nb); uint8_t * restrict pb = ((uint8_t *)y + 0*bs + 2*sizeof(float));
uint8_t pp[QK/2]; uint8_t pp[QK/2];
@ -627,8 +628,10 @@ void quantize_row_q4_1(const float * restrict x, void * restrict y, int k) {
const float d = (max - min) / ((1 << 4) - 1); const float d = (max - min) / ((1 << 4) - 1);
const float id = d ? 1.0f/d : 0.0f; const float id = d ? 1.0f/d : 0.0f;
pm[i] = min; *(float *)pm = min;
pd[i] = d; *(float *)pd = d;
pm += bs;
pd += bs;
for (int l = 0; l < QK; l += 2) { for (int l = 0; l < QK; l += 2) {
const float v0 = (x[i*QK + l + 0] - min)*id; const float v0 = (x[i*QK + l + 0] - min)*id;
@ -643,7 +646,8 @@ void quantize_row_q4_1(const float * restrict x, void * restrict y, int k) {
pp[l/2] = vi0 | (vi1 << 4); pp[l/2] = vi0 | (vi1 << 4);
} }
memcpy(pb + i*QK/2, pp, sizeof(pp)); memcpy(pb, pp, sizeof(pp));
pb += bs;
} }
} }
@ -687,16 +691,17 @@ void dequantize_row_q4_1(const void * restrict x, float * restrict y, int k) {
assert(k % QK == 0); assert(k % QK == 0);
const int nb = k / QK; const int nb = k / QK;
const size_t bs = 2*sizeof(float) + QK/2;
const float * restrict pm = (const float *) (x); const uint8_t * restrict pd = ((const uint8_t *)x + 0*bs);
const float * restrict pd = (const float *) (pm + nb); const uint8_t * restrict pm = ((const uint8_t *)x + 0*bs + sizeof(float));
const uint8_t * restrict pb = (const uint8_t *) (pd + nb); const uint8_t * restrict pb = ((const uint8_t *)x + 0*bs + 2*sizeof(float));
for (int i = 0; i < nb; i++) { for (int i = 0; i < nb; i++) {
const float m = pm[i]; const float d = *(const float *) (pd + i*bs);
const float d = pd[i]; const float m = *(const float *) (pm + i*bs);
const uint8_t * restrict pp = pb + i*QK/2; const uint8_t * restrict pp = pb + i*bs;
for (int l = 0; l < QK; l += 2) { for (int l = 0; l < QK; l += 2) {
const uint8_t vi = pp[l/2]; const uint8_t vi = pp[l/2];
@ -1584,28 +1589,109 @@ inline static void ggml_vec_dot_q4_0(const int n, float * restrict s, const void
inline static void ggml_vec_dot_q4_1(const int n, float * restrict s, const void * restrict x, const void * restrict y) { inline static void ggml_vec_dot_q4_1(const int n, float * restrict s, const void * restrict x, const void * restrict y) {
const int nb = n / QK; const int nb = n / QK;
const float * restrict pm0 = (const float *) x; const size_t bs = 2*sizeof(float) + QK/2;
const float * restrict pm1 = (const float *) y;
const float * restrict pd0 = (const float *) (pm0 + nb); const uint8_t * restrict pd0 = ((const uint8_t *)x + 0*bs);
const float * restrict pd1 = (const float *) (pm1 + nb); const uint8_t * restrict pd1 = ((const uint8_t *)y + 0*bs);
const uint8_t * restrict pb0 = (const uint8_t *) (pd0 + nb); const uint8_t * restrict pm0 = ((const uint8_t *)x + 0*bs + sizeof(float));
const uint8_t * restrict pb1 = (const uint8_t *) (pd1 + nb); const uint8_t * restrict pm1 = ((const uint8_t *)y + 0*bs + sizeof(float));
const uint8_t * restrict pb0 = ((const uint8_t *)x + 0*bs + 2*sizeof(float));
const uint8_t * restrict pb1 = ((const uint8_t *)y + 0*bs + 2*sizeof(float));
float sumf = 0.0; float sumf = 0.0;
#if 1 #if defined(__AVX2__)
#if QK == 32
// Initialize accumulator with zeros
__m256 acc = _mm256_setzero_ps();
// Accumulator for constant offsets
float acc_offset = 0.0f;
// Main loop
for (int i = 0; i < nb; ++i) {
const float * m0 = (const float *) (pm0 + i*bs);
const float * m1 = (const float *) (pm1 + i*bs);
const float * d0 = (const float *) (pd0 + i*bs);
const float * d1 = (const float *) (pd1 + i*bs);
const uint8_t * restrict p0 = pb0 + i*bs;
const uint8_t * restrict p1 = pb1 + i*bs;
const __m256 d0v = _mm256_broadcast_ss( d0 );
const __m256 d1v = _mm256_broadcast_ss( d1 );
const __m256 m0v = _mm256_broadcast_ss( m0 );
const __m256 m1v = _mm256_broadcast_ss( m1 );
// Compute combined scale for the block
const __m256 scale_01 = _mm256_mul_ps( d0v, d1v );
// Compute cross scales for the block
const __m256 scale_0 = _mm256_mul_ps( d0v, m1v );
const __m256 scale_1 = _mm256_mul_ps( m0v, d1v );
const __m256 cross_scales = _mm256_blend_ps( scale_0, scale_1, 0b10101010 );
// Load 16 bytes, and unpack 4 bit fields into bytes, making 32 bytes
__m256i bx = bytesFromNibbles( p0 );
__m256i by = bytesFromNibbles( p1 );
// Now we have a vector with bytes in [ 0 .. 15 ] interval.
// Sign-extend first 16 signed bytes into int16_t
__m256i x16 = _mm256_cvtepi8_epi16( _mm256_castsi256_si128( bx ) );
__m256i y16 = _mm256_cvtepi8_epi16( _mm256_castsi256_si128( by ) );
// Compute products of int16_t integers, add pairwise
__m256i i32 = _mm256_madd_epi16( x16, y16 );
// Sign-extend last 16 signed bytes into int16_t vectors
__m256i x16_h = _mm256_cvtepi8_epi16( _mm256_extracti128_si256( bx, 1 ) );
__m256i y16_h = _mm256_cvtepi8_epi16( _mm256_extracti128_si256( by, 1 ) );
// Accumulate products of int16_t integers
i32 = _mm256_add_epi32( i32, _mm256_madd_epi16( x16_h, y16_h ) );
// compute sums of unsigned bytes in bx, by in blocks of 8.
// This results in a layout like X100 0000 X200 0000 X300 0000 X400 0000,
// which we then interleave as X100 Y100 X200 Y200 X300 Y300 X400 Y400.
// so if we then cast to 8 singles, we get 8 floats like [ x0_7, y0_7, x8_15, y8_15, x16_23, y16_23, x24_31, y24_31 ]
__m256i xsumi = _mm256_sad_epu8( bx, _mm256_setzero_si256() );
__m256i ysumi = _mm256_sad_epu8( by, _mm256_setzero_si256() );
__m256i sumsi = _mm256_or_si256( xsumi, _mm256_slli_si256( ysumi, 4 ) );
__m256 sums = _mm256_cvtepi32_ps( sumsi );
// Convert int32_t to float
__m256 p = _mm256_cvtepi32_ps( i32 );
// Apply the scale, and accumulate
// acc += d0*d1*x*y + d0*m1*x + d1*m0*y
acc = _mm256_fmadd_ps( scale_01, p, acc );
acc = _mm256_fmadd_ps( cross_scales, sums, acc );
// acc_offset += m0*m1 (for each entry in the block)
acc_offset += (*m0)*(*m1);
}
// Return horizontal sum of the acc vector
__m128 res = _mm256_extractf128_ps( acc, 1 );
res = _mm_add_ps( res, _mm256_castps256_ps128( acc ) );
res = _mm_add_ps( res, _mm_movehl_ps( res, res ) );
res = _mm_add_ss( res, _mm_movehdup_ps( res ) );
sumf = _mm_cvtss_f32( res ) + acc_offset * QK;
#else
#error "not implemented for QK"
#endif
#else
// scalar // scalar
for (int i = 0; i < nb; i++) { for (int i = 0; i < nb; i++) {
const float m0 = pm0[i]; const float m0 = *(const float *) (pm0 + i*bs);
const float m1 = pm1[i]; const float m1 = *(const float *) (pm1 + i*bs);
const float d0 = pd0[i]; const float d0 = *(const float *) (pd0 + i*bs);
const float d1 = pd1[i]; const float d1 = *(const float *) (pd1 + i*bs);
const uint8_t * restrict p0 = pb0 + i*QK/2; const uint8_t * restrict p0 = pb0 + i*bs;
const uint8_t * restrict p1 = pb1 + i*QK/2; const uint8_t * restrict p1 = pb1 + i*bs;
for (int j = 0; j < QK/2; j++) { for (int j = 0; j < QK/2; j++) {
const uint8_t v0 = p0[j]; const uint8_t v0 = p0[j];
@ -1839,16 +1925,17 @@ inline static void ggml_vec_mad_q4_1(const int n, float * restrict y, void * res
assert(n % QK == 0); assert(n % QK == 0);
const int nb = n / QK; const int nb = n / QK;
const size_t bs = 2*sizeof(float) + QK/2;
const float * restrict pm = (const float *) (x); const uint8_t * restrict pd = ((const uint8_t *)x + 0*bs);
const float * restrict pd = (const float *) (pm + nb); const uint8_t * restrict pm = ((const uint8_t *)x + 0*bs + sizeof(float));
const uint8_t * restrict pb = (const uint8_t *) (pd + nb); const uint8_t * restrict pb = ((const uint8_t *)x + 0*bs + 2*sizeof(float));
for (int i = 0; i < nb; i++) { for (int i = 0; i < nb; i++) {
const float m = pm[i]; const float d = *(const float *) (pd + i*bs);
const float d = pd[i]; const float m = *(const float *) (pm + i*bs);
const uint8_t * restrict pp = pb + i*QK/2; const uint8_t * restrict pp = pb + i*bs;
for (int l = 0; l < QK; l += 2) { for (int l = 0; l < QK; l += 2) {
const uint8_t vi = pp[l/2]; const uint8_t vi = pp[l/2];

View file

@ -489,7 +489,8 @@ size_t ggml_quantize_q4_0(float * src, void * dst, int n, int k, int qk, int64_t
size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t * hist) { size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t * hist) {
const int nb = k / qk; const int nb = k / qk;
const size_t row_size = nb*(2*sizeof(float) + sizeof(uint8_t)*qk/2); const size_t bs = (2*sizeof(float) + sizeof(uint8_t)*qk/2);
const size_t row_size = nb*bs;
assert(k % qk == 0); assert(k % qk == 0);
@ -498,10 +499,10 @@ size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t
char * pdst = (char *) dst; char * pdst = (char *) dst;
for (int j = 0; j < n; j += k) { for (int j = 0; j < n; j += k) {
float * pm = (float *) (pdst + (j/k)*row_size); uint8_t * pd = (uint8_t *) (pdst + (j/k)*row_size + 0*bs);
float * pd = (float *) (pm + nb); uint8_t * pm = (uint8_t *) (pdst + (j/k)*row_size + 0*bs + sizeof(float));
uint8_t * pb = (uint8_t *) (pd + nb); uint8_t * pb = (uint8_t *) (pdst + (j/k)*row_size + 0*bs + 2*sizeof(float));
//printf("n = %d, k = %d, nb = %d, row_size = %d, j = %d, pm = %p, pd = %p, pb = %p\n", n, k, nb, row_size, j, pm, pd, pb); //printf("n = %d, k = %d, nb = %d, row_size = %d, j = %d, pm = %p, pd = %p, pb = %p\n", n, k, nb, row_size, j, pm, pd, pb);
@ -519,8 +520,10 @@ size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t
const float d = (max - min) / ((1 << 4) - 1); const float d = (max - min) / ((1 << 4) - 1);
const float id = d ? 1.0f/d : 0.0f; const float id = d ? 1.0f/d : 0.0f;
pm[i] = min; *(float *) pd = d;
pd[i] = d; *(float *) pm = min;
pd += bs;
pm += bs;
for (int l = 0; l < qk; l += 2) { for (int l = 0; l < qk; l += 2) {
const float v0 = (src[j + i*qk + l + 0] - min)*id; const float v0 = (src[j + i*qk + l + 0] - min)*id;
@ -538,7 +541,8 @@ size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t
pp[l/2] = vi0 | (vi1 << 4); pp[l/2] = vi0 | (vi1 << 4);
} }
memcpy(pb + i*qk/2, pp, pp_size); memcpy(pb, pp, pp_size);
pb += bs;
} }
} }
} }