root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-30 21:34:36 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Change this back to be in agreement with metal and our previous softmax kernel.

Browse Source
This commit is contained in:
Adam Treat 2023-10-03 13:30:23 -04:00 committed by cebtenzzre
parent 6ac39752bf
commit de589ced7c
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-vulkan.cpp
View File

@ -785,7 +785,7 @@ void ggml_vk_soft_max(kp::Sequence& seq,
std::shared_ptr<kp::Algorithm> s_algo = nullptr;
if (!komputeManager()->hasAlgorithm(__func__)) {
const uint32_t local_x = ggml_vk_current_device().subgroupSize * 2;
const uint32_t local_x = ggml_vk_current_device().subgroupSize;
s_algo = komputeManager()->algorithm<uint32_t, PushConstants>(__func__, s_kompute_context->pool.get(), {in, out}, spirv, {unsigned(ne01), unsigned(ne02), unsigned(ne03)}, {local_x}, {pushConsts});
} else {
s_algo = komputeManager()->getAlgorithm(__func__);