Repository navigation
Conversation
⚡ Performance Regression Check Results✅ Performance Check Passed (interpreter)Performance Benchmark Results (threshold: 25%)
Summary: 194 benchmarks, 0 regressions ✅ Performance Check Passed (multipass)Performance Benchmark Results (threshold: 25%)
Summary: 194 benchmarks, 0 regressions |
abmcar
marked this pull request as ready for review
October 10, 2026 10:04
joinVirtRegs built both JoinVals, whose constructors size arrays by the interval's value count, before asking isHighCostLiveInterval whether to give up. On EVM functions, whose stack-size and stack-top intervals hold thousands of values, most of those attempts are abandoned. Check first; the coalescer's decisions and the generated code are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The greedy allocator treated the instance pointer, an EVM function's only parameter and read throughout it, like any other long interval: on large contracts it split, spilled and revisited it until that work dominated compile time. Mark its interval unspillable before greedy allocation, via an opt-in compileMIRToCgIR flag that only the EVM compiler sets. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
abmcar
force-pushed
the
perf/evm-keep-instance-pointer-in-register
branch
from
October 11, 2026 07:06
6cf2c08 to
92413f2
Compare
abmcar
changed the title
6 of 16 tasks
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
1. Does this PR affect any open issues?(Y/N) and add issue references (e.g. "fix #123", "re #123".):
2. What is the scope of this PR (e.g. component or file name):
src/compiler: the register coalescer,compileMIRToCgIR, and the EVM compiler.3. Provide a description of the PR(e.g. more details, effects, motivations or doc link):
Register allocation dominates compile time on large EVM contracts. Two commits:
isHighCostLiveIntervalbefore building its twoJoinValsinstead of after. Same decisions, same code.compileMIRToCgIR; WebAssembly compilation is unchanged.Execution time of the generated code on
evmone-benchis unchanged within noise (geometric mean 0.998 on themainset); individual benchmarks move by up to about 2% either way.Change doc:
docs/changes/2026-10-11-evm-regalloc-compile-time4. Are there any breaking changes?(Y/N) and describe the breaking changes(e.g. more details, motivations or doc link):
5. Are there test cases for these changes?(Y/N) select and add more details, references or doc links:
No new tests.
ctest(default and CI multipass arguments) andtools/run_evm_tests.py(multipass, interpreter) pass. Benchmark stats are above.6. Release note
🤖 Generated with Claude Code