Skip to content

Fix llama.cpp ROCm image: ldconfig for runtime library resolution - #16

Merged
Aaronontheweb merged 1 commit into
masterfrom
fix/amd-rocm-ldconfig
Aug 14, 2026
Merged

Aaronontheweb merged 1 commit into
masterfrom
fix/amd-rocm-ldconfig

Conversation

@Aaronontheweb

Copy link
Copy Markdown
Member

Fourth candidate build compiled cleanly, installed, and passed the AVX-512 guard (no zmm opcodes in libggml-cpu), then failed only the smoke test:

llama-server: error while loading shared libraries: libllama-server-impl.so: cannot open shared object file

The installed binary in /opt/llama.cpp/bin can't find its .so files in /opt/llama.cpp/lib. Register that dir with ldconfig so llama-server resolves its own shared objects, without touching LD_LIBRARY_PATH (leaves ROCm's resolution intact).

…linking

Fourth candidate build compiled, installed, and PASSED the AVX-512 guard
(no zmm opcodes in libggml-cpu), then failed the smoke test: llama-server
could not load libllama-server-impl.so because /opt/llama.cpp/lib is not on
the default library search path. Register it via ldconfig so llama-server
resolves its own shared objects without touching LD_LIBRARY_PATH (leaving
ROCm's own resolution intact).
@Aaronontheweb
Aaronontheweb merged commit 98e4410 into master Aug 14, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant