ci(gfx11): cache HIP builds with ggml-org/ccache-actio#45
Open
roberteg16 wants to merge 3 commits into
Open
Conversation
The gfx11 ROCm workflow recompiles all 7 RDNA3/3.5 arches from scratch on every run. Add the repo-standard ggml-org/ccache-action before the build so object output persists across runs via actions/cache. llama.cpp's CMake auto-wires ccache as the compile launcher (GGML_CCACHE defaults ON), so no cmake flags are needed. Cache writes are gated to pushes on gfx11 so PRs read the shared store without thrashing it. Co-Authored-By: Claude Opus 4 (1M context) <noreply@anthropic.com>
roberteg16
changed the title
Co-Authored-By: Claude Opus 4 (1M context) <noreply@anthropic.com>
roberteg16
changed the title
mgehre-amd
reviewed
Jul 13, 2026
gfx11 saves the cache only on push (infrequent, plus the daily nightly), so a 1d eviction window prunes still-useful objects between runs. 7d keeps hot objects across the branch's lower push cadence. Co-Authored-By: Claude Opus 4 (1M context) <noreply@anthropic.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
build-gfx11-rocmworkflow recompiles all 7 RDNA3/3.5 arches from scratch on every push/PR/dispatch — no build caching today.ggml-org/ccache-action@v1.2.21(already used by the CUDA/HIP/SYCL/Vulkan/release workflows) before the build step, so object output persists across runs via GitHub'sactions/cache.GGML_CCACHEis ON (default) and ccache is onPATH(ggml/src/CMakeLists.txt), wrapping the explicit/opt/rocm/llvm/bin/clang.Behavior
gfx11(save: github.event_name == 'push' && github.ref == 'refs/heads/gfx11'), so PRs read the shared store without thrashing it.evict-old-files: 1dbounds store growth.Test plan
ccachestep restoring/creating the cache and reports ccache stats at build end.ccache found, compilation results will be cached.🤖 Generated with Claude Code