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

cuda : remove nchannels_x argument from mul_mat_vec_nc_f16_f32 (#2028)

Browse source 
- Not used
This commit is contained in:
Salvador E. Tropea 2023-06-28 14:27:31 -03:00 committed by GitHub
parent 6432aabb6d
commit 5b351e94d0
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
ggml-cuda.cu
View file

@ -1292,7 +1292,7 @@ static __global__ void mul_mat_p021_f16_f32(const void * vx, const float * y, fl
static __global__ void mul_mat_vec_nc_f16_f32( // nc == non-contiguous
const void * vx, const float * y, float * dst, const int ncols_x, const int nrows_x,
const int row_stride_x, const int nchannels_x, const int channel_stride_x) {
const int row_stride_x, const int channel_stride_x) {
const half * x = (const half *) vx;
@ -1698,7 +1698,7 @@ static void ggml_mul_mat_vec_nc_f16_f32_cuda(
const dim3 block_nums(1, nrows_x, nchannels_x);
const dim3 block_dims(WARP_SIZE, 1, 1);
mul_mat_vec_nc_f16_f32<<<block_nums, block_dims, 0, stream>>>
(vx, y, dst, ncols_x, nrows_x, row_stride_x, nchannels_x, channel_stride_x);
(vx, y, dst, ncols_x, nrows_x, row_stride_x, channel_stride_x);
}
static void ggml_cpy_f32_f32_cuda(