ci: disable sccache for Windows ROCm native runtime build - #1087
Merged
Conversation
sccache + hipcc is broken on Windows: HIP compiles one host action plus one device action per --offload-arch and bundles them, and combined with the long arch-derived build directory (build-stage-abi-dynamic-rocm-gfx90a_..._gfx1201) the object paths exceed MAX_PATH. sccache then fails to persist its temp file (os error 3) and the build aborts. This only hits the Windows ROCm native-runtime row: Windows CUDA/Vulkan native runtimes and the Windows ROCm release bundle (shorter build dir) all pass with the same sccache config. Set LLAMA_STAGE_USE_SCCACHE=0 for that row only, which build-llama.sh already honors by omitting the HIP compiler launcher. CUDA and Vulkan keep sccache.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughThe Windows GPU release workflow now disables sccache for the ROCm matrix entry while defaulting sccache to enabled for other native runtime packaging rows. ChangesWindows GPU release configuration
Estimated code review effort: 1 (Trivial) | ~5 minutes Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
michaelneale
added a commit
that referenced
this pull request
Jul 27, 2026
* origin/main: Fix Metal small-batch matmul parity for GLM verification (#1078) Handle K-only transposed KV page import and export (#1084) Refresh llama.cpp upstream patch queue (#1085) chore: improve embedded native-runtime compatibility guidance (#1043) fix(console-ui): chat transcript snapping during live status updates (#1083) ci: bump Linux CUDA slim container to gha-convention base runner image fix: record activation cache prefix identities (#1041) fix: read-only model download caches (#1042) ci: disable sccache for Windows ROCm native runtime build (#1087) ci: fix v0.74 release GPU builds (sccache disk-only + force_hosted_runners) (#1086) Make release sccache failures non-fatal (#1079) Keep client-only nodes out of model election (#1074) # Conflicts: # crates/mesh-llm-host-runtime/src/runtime/auto_join.rs # crates/mesh-llm-host-runtime/src/runtime/tests/auto_join.rs # third_party/llama.cpp/patches/0004-Add-lanes-external-media-and-chat-grammar-support.patch
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
1 participant
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.
What this fixes
The final blocker for a complete
v0.74.0GPU release. rc4 got every Linux CUDA/ROCm/Vulkan and Windows CUDA/Vulkan lane green (node24 + sccache-token-expiry both fixed in #1086), but one lane failed deterministically:Build native runtime Windows x86_64 ROCm→Root cause
sccache + hipcc is broken on Windows for this build. HIP compiles one host action plus one device action per
--offload-arch(9 arches here) and bundles them; combined with the long arch-derived build directory (build-stage-abi-dynamic-rocm-gfx90a_..._gfx1201), object paths exceed WindowsMAX_PATH, and sccache fails the final atomic temp-file persist (os error 3).Evidence this is ROCm-native-specific, not a general path/sccache/shell problem:
build-stage-abi-rocm(short)build-stage-abi-dynamic-rocm-<9 arches>(long)sccache's HIP support is known-fragile (PyTorch disables ccache/sccache for HIP for similar reasons).
Fix
Set
LLAMA_STAGE_USE_SCCACHE=0for the Windows ROCm native-runtime matrix row only.scripts/build-llama.shalready honors this by omitting the HIP compiler launcher, so the build proceeds without the caching layer that can't handle these paths. CUDA and Vulkan keep sccache unchanged (defaulted to1viamatrix.use_sccache || '1').Validation
actionlint -config-file .github/actionlint.yaml— passgit diff --check— cleancargo run -p xtask -- repo-consistency release-targets— passReal end-to-end proof is the next RC (rc5) once merged.
Rollback
Summary by CodeRabbit