]> git.djapps.eu Git - pkg/ggml/sources/whisper.cpp/commit
opencl: add q4_1 MoE for Adreno (llama/22856)
authorShawn Gu <redacted>
Mon, 11 May 2026 18:57:26 +0000 (11:57 -0700)
committerGeorgi Gerganov <redacted>
Thu, 14 May 2026 18:26:48 +0000 (21:26 +0300)
commitea4652c42704fc298605fb639dedc40937845c78
tree4d5e18dbdcf503de9e4bc34bd8ef0535d37aeb3f
parent287f637fb15f3c59d0f579f86c4b59d3ebfbc443
opencl: add q4_1 MoE for Adreno (llama/22856)

* Q4_1 MoE CLC pass sanity check

* remove unnecessary code

* opencl: remove unnecessary asserts and reformat

* opencl: fix supports_op for q4_1 moe

* q4_1 moe is supported by Adreno with certain shapes

---------

Co-authored-by: Li He <redacted>
ggml/src/ggml-opencl/CMakeLists.txt
ggml/src/ggml-opencl/ggml-opencl.cpp
ggml/src/ggml-opencl/kernels/cvt.cl
ggml/src/ggml-opencl/kernels/gemm_moe_q4_1_f32_ns.cl [new file with mode: 0644]
ggml/src/ggml-opencl/kernels/gemv_moe_q4_1_f32_ns.cl [new file with mode: 0644]