]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 7fee288)
HIP: Only call rocblas_initialize on rocblas versions with the multiple instantation...
authorNikita Sarychev <redacted>
Tue, 28 Jan 2025 15:42:20 +0000 (07:42 -0800)
committerGitHub <redacted>
Tue, 28 Jan 2025 15:42:20 +0000 (16:42 +0100)
commitcae9fb4361138b937464524eed907328731b81f6
tree803e3243414b9af9cfbf4da2cc9980c94e8231f6tree
parent7fee2889e6565830631fbe76d47ef85cf8fd946acommit | diff
HIP: Only call rocblas_initialize on rocblas versions with the multiple instantation bug (#11080)

This disables the workaround on rocblas fixed versions (>=4.0.0) to eliminate the runtime cost and unnecessary VRAM allocation of loading all tensile objects.
ggml/src/ggml-cuda/ggml-cuda.cu diff | blob | history
Packaging of ggml-org/llama.cpp