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

Fix F32 matmul

Browse source
This commit is contained in:
0cc4m 2023-07-16 07:55:53 +02:00
parent f58fa51fd0
commit 931a8921de
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
ggml-vulkan.cpp
View file

@ -1241,7 +1241,7 @@ static void ggml_vk_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
if (src0->backend == GGML_BACKEND_GPU) { if (src0->backend == GGML_BACKEND_GPU) {
d_X = *(vk_buffer*) src0->data; d_X = *(vk_buffer*) src0->data;
} else { } else {
ggml_vk_pool_malloc(ggml_type_size(src0->type) * x_ne, &d_X, 0); ggml_vk_pool_malloc(sizeof(float) * x_ne, &d_X, 0);
} }
ggml_vk_pool_malloc(sizeof(float) * y_ne, &d_Y, 0); ggml_vk_pool_malloc(sizeof(float) * y_ne, &d_Y, 0);
ggml_vk_pool_malloc(sizeof(float) * d_ne * split_k, &d_D, 0); ggml_vk_pool_malloc(sizeof(float) * d_ne * split_k, &d_D, 0);
@ -1253,7 +1253,7 @@ static void ggml_vk_mul_mat_f32(const ggml_tensor * src0, const ggml_tensor * sr
vk::Semaphore s_it_x; vk::Semaphore s_it_x;
vk::Semaphore s_it_y; vk::Semaphore s_it_y;
const bool load_x = src0->backend != GGML_BACKEND_CPU; const bool load_x = src0->backend != GGML_BACKEND_GPU;
for (int64_t i03 = 0; i03 < ne03; i03++) { for (int64_t i03 = 0; i03 < ne03; i03++) {
for (int64_t i02 = 0; i02 < ne02; i02++) { for (int64_t i02 = 0; i02 < ne02; i02++) {