]> git.djapps.eu Git - pkg/ggml/sources/whisper.cpp/commit
vulkan: Implement topk_moe fused shader, ported from CUDA (llama/16641)
authorJeff Bolz <redacted>
Sat, 18 Oct 2025 10:22:57 +0000 (05:22 -0500)
committerGeorgi Gerganov <redacted>
Wed, 22 Oct 2025 09:58:11 +0000 (12:58 +0300)
commit414901a42c4ac9998615d45ee4e0f6cfe3064377
tree6aaa1d931ae8be5caea9ca82f9d9f9529efc098d
parent08345f15ece9bdc528770596c08b48144082e933
vulkan: Implement topk_moe fused shader, ported from CUDA (llama/16641)

This is similar to the CUDA shader from #16130, but doesn't use shared memory
and handles different subgroup sizes.
ggml/src/ggml-impl.h
ggml/src/ggml-vulkan/ggml-vulkan.cpp
ggml/src/ggml-vulkan/vulkan-shaders/topk_moe.comp [new file with mode: 0644]
ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp