Hongqiang Wang 3af988fabc opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (#26431) ...

* opencl: fold the gpt-oss MoE bias adds into swiglu_oai

Default on, opt out with GGML_OPENCL_FUSE_MOE_BIAS_GLU=0.

* opencl: fold the MoE down-projection bias into the combine

Default on, opt out with GGML_OPENCL_FUSE_MOE_BIAS_COMBINE=0.

2026-08-21 14:24:33 -07:00

..

kernels

opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (#26431)

2026-08-21 14:24:33 -07:00

cl-program-cache.cpp

opencl: cache compiled cl_program binaries on disk (#26050)

2026-07-24 08:14:33 -07:00

cl-program-cache.h

opencl: cache compiled cl_program binaries on disk (#26050)

2026-07-24 08:14:33 -07:00

CMakeLists.txt

opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (#26431)

2026-08-21 14:24:33 -07:00

fa_tune.h

opencl: general flash attention decode performance optimizations (#25366)

2026-07-06 19:57:52 -07:00

ggml-opencl.cpp

opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (#26431)

2026-08-21 14:24:33 -07:00

libdl.h

opencl: allow loading precompiled binary kernels from library (#23042)

2026-07-01 10:29:22 -07:00