vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix QKK_64

Browse source
This commit is contained in:
Henrik Forstén 2024-01-02 15:44:45 +02:00
parent 4b06507172
commit 7e03872f1e
1 changed files with 57 additions and 35 deletions
Download patch file Download diff file Expand all files Collapse all files

92
ggml-quants.c
View file

@ -5,7 +5,6 @@
#include <string.h> #include <string.h>
#include <assert.h> #include <assert.h>
#include <float.h> #include <float.h>
#include <stdio.h>
#ifdef __ARM_NEON #ifdef __ARM_NEON
@ -2041,6 +2040,29 @@ void quantize_row_q4_K_reference(const float * restrict x, block_q4_K * restrict
} }
} }
#else #else
float weights[32];
float mins[QK_K/32];
float scales[QK_K/32];
uint8_t Laux[32];
float max_scale = 0; // as we are deducting the min, scales are always positive
float max_min = 0;
for (int j = 0; j < QK_K/32; ++j) {
//scales[j] = make_qkx1_quants(32, 15, x + 32*j, L + 32*j, &mins[j], 9, 0.5f);
float sum_x2 = 0;
for (int l = 0; l < 32; ++l) sum_x2 += x[32*j + l] * x[32*j + l];
float av_x = sqrtf(sum_x2/32);
for (int l = 0; l < 32; ++l) weights[l] = av_x + fabsf(x[32*j + l]);
scales[j] = make_qkx2_quants(32, 15, x + 32*j, weights, L + 32*j, &mins[j], Laux, -1.f, 0.1f, 20, false);
float scale = scales[j];
if (scale > max_scale) {
max_scale = scale;
}
float min = mins[j];
if (min > max_min) {
max_min = min;
}
}
const float s_factor = 15.f; const float s_factor = 15.f;
float inv_scale = max_scale > 0 ? s_factor/max_scale : 0.f; float inv_scale = max_scale > 0 ? s_factor/max_scale : 0.f;
float inv_min = max_min > 0 ? s_factor/max_min : 0.f; float inv_min = max_min > 0 ? s_factor/max_min : 0.f;
@ -2156,43 +2178,43 @@ void quantize_row_q5_K_reference(const float * restrict x, block_q5_K * restrict
for (int i = 0; i < nb; i++) { for (int i = 0; i < nb; i++) {
#if QK_K == 256 #if QK_K == 256
int block_scales[QK_K/32]; int block_scales[QK_K/32];
int block_mins[QK_K/32]; int block_mins[QK_K/32];
quantize_q_k_1(x, 5, 6, 32, block_scales, block_mins, L, &y[i].d, &y[i].dmin); quantize_q_k_1(x, 5, 6, 32, block_scales, block_mins, L, &y[i].d, &y[i].dmin);
for (int j = 0; j < QK_K/32; ++j) { for (int j = 0; j < QK_K/32; ++j) {
int ls = block_scales[j]; int ls = block_scales[j];
int lm = block_mins[j]; int lm = block_mins[j];
if (j < 4) { if (j < 4) {
y[i].scales[j] = ls; y[i].scales[j] = ls;
y[i].scales[j+4] = lm; y[i].scales[j+4] = lm;
} else { } else {
y[i].scales[j+4] = (ls & 0xF) | ((lm & 0xF) << 4); y[i].scales[j+4] = (ls & 0xF) | ((lm & 0xF) << 4);
y[i].scales[j-4] |= ((ls >> 4) << 6); y[i].scales[j-4] |= ((ls >> 4) << 6);
y[i].scales[j-0] |= ((lm >> 4) << 6); y[i].scales[j-0] |= ((lm >> 4) << 6);
}
}
uint8_t * restrict qh = y[i].qh;
uint8_t * restrict ql = y[i].qs;
memset(qh, 0, QK_K/8);
uint8_t m1 = 1, m2 = 2;
for (int n = 0; n < QK_K; n += 64) {
for (int j = 0; j < 32; ++j) {
int l1 = L[n + j];
if (l1 > 15) {
l1 -= 16; qh[j] |= m1;
} }
int l2 = L[n + j + 32];
if (l2 > 15) {
l2 -= 16; qh[j] |= m2;
}
ql[j] = l1 | (l2 << 4);
} }
m1 <<= 2; m2 <<= 2;
ql += 32; uint8_t * restrict qh = y[i].qh;
} uint8_t * restrict ql = y[i].qs;
memset(qh, 0, QK_K/8);
uint8_t m1 = 1, m2 = 2;
for (int n = 0; n < QK_K; n += 64) {
for (int j = 0; j < 32; ++j) {
int l1 = L[n + j];
if (l1 > 15) {
l1 -= 16; qh[j] |= m1;
}
int l2 = L[n + j + 32];
if (l2 > 15) {
l2 -= 16; qh[j] |= m2;
}
ql[j] = l1 | (l2 << 4);
}
m1 <<= 2; m2 <<= 2;
ql += 32;
}
#else #else
float max_scale = 0, amax = 0; float max_scale = 0, amax = 0;
for (int j = 0; j < QK_K/16; ++j) { for (int j = 0; j < QK_K/16; ++j) {