]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
opencl: add kernel to handle mat mul in attention to improve encoding speed (llama...
authorshaofeiqi <redacted>
Sun, 16 Nov 2025 01:33:10 +0000 (17:33 -0800)
committerGeorgi Gerganov <redacted>
Mon, 17 Nov 2025 13:34:43 +0000 (15:34 +0200)
commit3ef123f1ab0bd23917c1d095b09b7a5b6c3d49d2
treeddc83cabc14ea64e88771d8d585bce157744f009
parent2d58a1baaeae1446ad875ae38ce6e6fd618369fd
opencl: add kernel to handle mat mul in attention to improve encoding speed (llama/17181)

* Add mul_mm_f16_f32_kq_kqv kernel

* Add ggml_cl_mul_mat_kq_kqv_adreno func

* fix whitespace

* remove unused variable

* remove redundant

* refactor and clean up

* remove trailing whitespace
src/ggml-opencl/CMakeLists.txt
src/ggml-opencl/ggml-opencl.cpp
src/ggml-opencl/kernels/mul_mm_f16_f32_kq_kqv.cl [new file with mode: 0644]