]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 0136966)
update HIP_UMA #7399 (#7414)
authorDjip007 <redacted>
Mon, 27 May 2024 23:40:47 +0000 (01:40 +0200)
committerGitHub <redacted>
Mon, 27 May 2024 23:40:47 +0000 (01:40 +0200)
commit852aafb163d32d5bad63c10bc323a02c28fec59d
tree48c9bf7791952889c83da7ee3085818d7a61fa25tree
parent0136966dafb452601c23f30395878d5a65ddc559commit | diff
update HIP_UMA #7399 (#7414)

* update HIP_UMA #7399

add use of hipMemAdviseSetCoarseGrain when LLAMA_HIP_UMA is enable.
- get x2 on prompte eval and x1.5 on token gen with rocm6.0 on ryzen 7940HX iGPU (780M/gfx1103)

* simplify code, more consistent style

---------

Co-authored-by: slaren <redacted>
ggml-cuda.cu diff | blob | history
ggml-cuda/common.cuh diff | blob | history
Packaging of ggml-org/llama.cpp