]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
overview / pkg / ggml / sources / ggml / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 0beb02b)
opencl: fix for small models (llama/11950)
authorlhez <redacted>
Mon, 24 Feb 2025 21:47:07 +0000 (13:47 -0800)
committerGeorgi Gerganov <redacted>
Tue, 25 Feb 2025 11:33:09 +0000 (13:33 +0200)
commitdd26b0f9eba9aafdc504fede5c2c2d8c8bc59fbb
tree807ed1ed3756e307ba9672991840fc6c9809a3cbtree
parent0beb02b81aa503bafdaebec74506fba7a433c7c3commit | diff
opencl: fix for small models (llama/11950)

* opencl: fix small shape gemv, remove unused extensions

* opencl: fix `transpose_16`, `dump_tensor`, enforce subgroup size

* opencl: fix for token length < 4

* opencl: use wave size of 64 for all Adreno GPUs

---------

Co-authored-by: Shawn Gu <redacted>
Co-authored-by: Skyler Szot <redacted>
src/ggml-opencl/ggml-opencl.cpp diff | blob | history
src/ggml-opencl/kernels/ggml-opencl.cl diff | blob | history
src/ggml-opencl/kernels/ggml-opencl_gemv_noshuffle.cl diff | blob | history
src/ggml-opencl/kernels/ggml-opencl_gemv_noshuffle_general.cl diff | blob | history
src/ggml-opencl/kernels/ggml-opencl_mul_mat_Ab_Bi_8x4.cl diff | blob | history
src/ggml-opencl/kernels/ggml-opencl_transpose_16.cl diff | blob | history
Packaging of ggml-org/ggml