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

Fixing merge conflicts

Browse source
This commit is contained in:
Iwan Kawrakow 2023-06-08 20:17:45 +03:00
parent 1c97f2919a
commit 5a46032f31
1 changed files with 6 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml-metal.metal
View file

@ -623,11 +623,17 @@ static void dequantize_row_q4_k(device const block_q4_k * x, device float * y, i
} }
static void dequantize_row_q6_k(device const block_q6_k * x, device float * y, int k) { static void dequantize_row_q6_k(device const block_q6_k * x, device float * y, int k) {
assert(k % QK_K == 0);
const int nb = k / QK_K;
for (int i = 0; i < nb; i++) {
device const uint8_t * ql = x[i].ql; device const uint8_t * ql = x[i].ql;
device const uint8_t * qh = x[i].qh; device const uint8_t * qh = x[i].qh;
device const int8_t * sc = x[i].scales; device const int8_t * sc = x[i].scales;
const float d = x[i].d;
for (int n = 0; n < QK_K; n += 128) { for (int n = 0; n < QK_K; n += 128) {
for (int l = 0; l < 32; ++l) { for (int l = 0; l < 32; ++l) {
int is = l/16; int is = l/16;