]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
overview / pkg / ggml / sources / ggml / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: cfa85c8)
OpenCL: add initial FA support (llama/14987)
authorrmatif <redacted>
Sat, 16 Aug 2025 08:05:55 +0000 (10:05 +0200)
committerGeorgi Gerganov <redacted>
Mon, 18 Aug 2025 16:15:25 +0000 (19:15 +0300)
commit6733d5c937c82dd31d765861472a68be1f8ea2ea
tree329b2650bbdfa37382fe524b6b912cb956744d49tree
parentcfa85c83104f4657e5d800780a38a83c604e8161commit | diff
OpenCL: add initial FA support (llama/14987)

* add F16/F16 fa support

* fix kernel init

* use mad instead of fma

* use inline function

* mark FA with sinks as unsupported for now

* add pragma unroll to loops
src/ggml-opencl/CMakeLists.txt diff | blob | history
src/ggml-opencl/ggml-opencl.cpp diff | blob | history
src/ggml-opencl/kernels/flash_attn_f16.cl [new file with mode: 0644] blob
src/ggml-opencl/kernels/flash_attn_f32.cl [new file with mode: 0644] blob
src/ggml-opencl/kernels/flash_attn_f32_f16.cl [new file with mode: 0644] blob
Packaging of ggml-org/ggml