]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 91c36c2)
vulkan: optimize and reenable split_k (#10637)
authorJeff Bolz <redacted>
Tue, 3 Dec 2024 19:29:54 +0000 (13:29 -0600)
committerGitHub <redacted>
Tue, 3 Dec 2024 19:29:54 +0000 (20:29 +0100)
commitcc98896db858df7aa40d0e16a505883ef196a482
treece3a756e20c4b9149087914c119b7bfcf405114etree
parent91c36c269bca75b2d08119c653512cd20b4ea2bacommit | diff
vulkan: optimize and reenable split_k (#10637)

Use vector loads when possible in mul_mat_split_k_reduce. Use split_k
when there aren't enough workgroups to fill the shaders.
ggml/src/ggml-vulkan/ggml-vulkan.cpp diff | blob | history
ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_split_k_reduce.comp diff | blob | history
Packaging of ggml-org/llama.cpp