nix: Added missing packages and options for ROCm build #15747
+28
−8
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
github-actions
bot
added
nix
SteelPh0enix
force-pushed
the
nix-rocm-tweaks
branch
5 times, most recently
from
452acd8 to
72b59e4
Compare
SteelPh0enix
force-pushed
the
nix-rocm-tweaks
branch
from
72b59e4 to
85542e8
Compare
github-actions
bot
added
the
ggml
SteelPh0enix
force-pushed
the
nix-rocm-tweaks
branch
5 times, most recently
from
207ada3 to
f794bd3
Compare
ericcurtin
approved these changes
Sep 5, 2025
ericcurtin
approved these changes
Sep 5, 2025
SteelPh0enix
force-pushed
the
nix-rocm-tweaks
branch
from
f794bd3 to
b468b4c
Compare
SteelPh0enix
force-pushed
the
nix-rocm-tweaks
branch
from
b468b4c to
6ddaef4
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, Nix flake for llama-cpp-rocm doesn't build due to missing
lldandobjcopyexecutables. This MR fixes it by adding them to package list.Additionally, this MR:
mpipackage depending on selected backendbuilsAllCudaFaQuantsparameter to enable building all flash attention quants for CUDA and ROCm builds.rocmUseWmmato add ROCWMMA_FATTN support, enabled by default.enableUmaparameter to enable UMA support for CUDA/ROCm platformsIn order to enable WMMA support, i had to add a CMake variable and tweak the build script to provide a way of setting the path to include directory manually - otherwise CMake could not find the headers and failed the build
I've tested building the flake with/without ROCm and MPI support and it works fine now and produces working llama-server.