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

Update ggml/src/ggml-cuda/dmmv.cu

Browse source 
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
This commit is contained in:
agray3 2024-10-10 15:47:11 +01:00 • committed by GitHub
parent d07dc44c63
commit d150c7e309
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml/src/ggml-cuda/dmmv.cu
View file

@ -480,10 +480,7 @@ static __global__ void dequantize_mul_mat_vec(const void * __restrict__ vx, cons
if ( y_offset == 1 ) {
// load 2 dfloats into register in a single instruction
const dfloat2 y_reg = *((dfloat2 *) &(y[iybs + iqs + j/qr]));
tmp += __hmul2(v, {
y_reg.x;
y_reg.y;
});
tmp += __hmul2(v, y_reg);
}
else {
tmp += __hmul2(v, {