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

metal : fix accuracy of dequantization kernels

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2023-12-13 18:32:52 +02:00
parent bc014485be
commit 694449f0e8
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
2 changed files with 15 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

26
ggml-metal.metal
View file

@ -3521,10 +3521,10 @@ void dequantize_q8_0(device const block_q8_0 *xb, short il, thread type4x4 & reg
template <typename type4x4>
void dequantize_q2_K(device const block_q2_K *xb, short il, thread type4x4 & reg) {
const half d = xb->d;
const half min = xb->dmin;
const float d = xb->d;
const float min = xb->dmin;
device const uint8_t * q = (device const uint8_t *)xb->qs;
half dl, ml;
float dl, ml;
uint8_t sc = xb->scales[il];
#if QK_K == 256
@ -3594,10 +3594,10 @@ void dequantize_q4_K(device const block_q4_K *xb, short il, thread type4x4 & reg
q = q + (il/4) * 32 + 16 * (il&1);
il = il & 3;
const uchar2 sc = get_scale_min_k4_just2(is, il/2, xb->scales);
const half d = il < 2 ? xb->d : xb->d / 16.h;
const half min = xb->dmin;
const half dl = d * sc[0];
const half ml = min * sc[1];
const float d = il < 2 ? xb->d : xb->d / 16.h;
const float min = xb->dmin;
const float dl = d * sc[0];
const float ml = min * sc[1];
#else
q = q + 16 * (il&1);
device const uint8_t * s = xb->scales;
@ -3624,13 +3624,13 @@ void dequantize_q5_K(device const block_q5_K *xb, short il, thread type4x4 & reg
uint8_t ul = 1 << (il/2);
il = il & 3;
const uchar2 sc = get_scale_min_k4_just2(is, il/2, xb->scales);
const half d = il < 2 ? xb->d : xb->d / 16.h;
const half min = xb->dmin;
const half dl = d * sc[0];
const half ml = min * sc[1];
const float d = il < 2 ? xb->d : xb->d / 16.h;
const float min = xb->dmin;
const float dl = d * sc[0];
const float ml = min * sc[1];
const ushort mask = il<2 ? 0x0F : 0xF0;
const half qh_val = il<2 ? 16.h : 256.h;
const ushort mask = il<2 ? 0x0F : 0xF0;
const float qh_val = il<2 ? 16.f : 256.f;
for (int i = 0; i < 16; ++i) {
reg[i/4][i%4] = dl * ((q[i] & mask) + (qh[i] & ul ? qh_val : 0)) - ml;
}

3
tests/test-backend-ops.cpp
View file

@ -432,9 +432,10 @@ struct test_case {
if (err > ud->max_err) {
printf("[%s] NMSE = %f ", ggml_op_desc(t1), err);
//for (int i = 0; i < f1.size(); i++) {
// printf("(%f, %f) ", f1[i], f2[i]);
// printf("%5d %9.6f %9.6f, diff = %9.6f\n", i, f1[i], f2[i], f1[i] - f2[i]);
//}
//printf("\n");
//exit(1);
ud->ok = false;
}
return true;