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

Fix q3_k

Browse source
This commit is contained in:
0cc4m 2023-06-13 08:25:32 +02:00
parent fc8c823f34
commit 2a972f3649
1 changed files with 4 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml-opencl.cpp
View file

@ -349,6 +349,7 @@ void vec_dot_q3_K(__global const struct block_q3_K* x, const int ib, const int i
const uint32_t kmask1 = 0x03030303;
const uint32_t kmask2 = 0x0f0f0f0f;
uint32_t aux[3];
uint32_t utmp[4];
int n = iqs/128;
@ -360,7 +361,9 @@ void vec_dot_q3_K(__global const struct block_q3_K* x, const int ib, const int i
__global const uint8_t * hm = x[ib].hmask + l;
const int8_t * s = (const int8_t *)utmp + 8*n;
__global const uint32_t* aux = (__global const uint32_t*) x[ib].scales;
aux[0] = x[ib].scales[0] | x[ib].scales[1] << 8 | x[ib].scales[2] << 16 | x[ib].scales[3] << 24;
aux[1] = x[ib].scales[4] | x[ib].scales[5] << 8 | x[ib].scales[6] << 16 | x[ib].scales[7] << 24;
aux[2] = x[ib].scales[8] | x[ib].scales[9] << 8 | x[ib].scales[10] << 16 | x[ib].scales[11] << 24;
utmp[3] = ((aux[1] >> 4) & kmask2) | (((aux[2] >> 6) & kmask1) << 4);
utmp[2] = ((aux[0] >> 4) & kmask2) | (((aux[2] >> 4) & kmask1) << 4);