]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: d8bd001)
OpenCL: Fix duplication of layers in VRAM and RAM, add GPU mul kernel (#1653)
author0cc4m <redacted>
Sun, 4 Jun 2023 06:12:05 +0000 (08:12 +0200)
committerGitHub <redacted>
Sun, 4 Jun 2023 06:12:05 +0000 (08:12 +0200)
commitdcb2ed48268e421baf25adc00d602dad0f415564
tree261ef84fe660d06fce90c58fc01a16ae0e69be52tree
parentd8bd0013e8768aaa3dc9cfc1ff01499419d5348ecommit | diff
OpenCL: Fix duplication of layers in VRAM and RAM, add GPU mul kernel (#1653)

* Use events instead of clFinish, where possible

* OpenCL: Don't load gpu layers into RAM, add mul_f32 kernel

* Reduce queueing overhead for contiguous tensors by using single mul kernel call

* Adapt to #1612 cl_mem malloc changes

* Reduce code duplication between cuda and opencl branches

* Improve implementation
ggml-opencl.cpp diff | blob | history
ggml-opencl.h diff | blob | history
ggml.c diff | blob | history
llama.cpp diff | blob | history
Packaging of ggml-org/llama.cpp