]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
opencl: add q4_K gemm and gemv kernels for Adreno (llama/20919)
authorshaofeiqi <redacted>
Mon, 30 Mar 2026 19:19:16 +0000 (12:19 -0700)
committerGeorgi Gerganov <redacted>
Wed, 1 Apr 2026 13:00:26 +0000 (16:00 +0300)
commitc0fe99a7611cc76f743f055b35c3302da0058ae2
tree56406f32049953b6dc99bceca97d18901c3a8432
parent9e9adf696bd0e7da80d5166a8a4c214b5f2027cb
opencl: add q4_K gemm and gemv kernels for Adreno (llama/20919)

* opencl: add q4_K gemm and gemv kernels for Adreno

* opencl: fix whitespace

* opencl: add workarounds for compiler bugs on older devices

* opencl: handle fp16 denorm on X Elite

* opencl: fix kernel build error

* opencl: fix whitespace

* opencl: make q4_K cvt kernels signature consistent

---------

Co-authored-by: Li He <redacted>
src/ggml-opencl/CMakeLists.txt
src/ggml-opencl/ggml-opencl.cpp
src/ggml-opencl/kernels/cvt.cl
src/ggml-opencl/kernels/gemm_noshuffle_q4_k_f32.cl [new file with mode: 0644]
src/ggml-opencl/kernels/gemv_noshuffle_q4_k_f32.cl [new file with mode: 0644]