]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: f5e96b3)
vulkan: optimizations for deepseek prompt processing (#14555)
authorJeff Bolz <redacted>
Sat, 12 Jul 2025 09:51:58 +0000 (04:51 -0500)
committerGitHub <redacted>
Sat, 12 Jul 2025 09:51:58 +0000 (11:51 +0200)
commit98197e5c98388470030d908f355ec5937dcccaaa
tree878b08a09a74d501684c6837940243d3ab24e34dtree
parentf5e96b368f1acc7f53c390001b936517c4d18999commit | diff
vulkan: optimizations for deepseek prompt processing (#14555)

* vulkan: allow unclamped loads in coopmat2 mul_mat_id shader

* vulkan: increase coopmat2 mul_mat_id tile size

* vulkan: optimize mat_mul_id row_ids search to batch loads, and port to coopmat1 path

* vulkan: use smaller FA row size when head size is large. applies to both scalar and CM2 paths (CM1 isn't used due to shared memory limits)
ggml/src/ggml-vulkan/ggml-vulkan.cpp diff | blob | history
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp diff | blob | history
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp diff | blob | history
Packaging of ggml-org/llama.cpp