test(vm): Improve instruction-counting VM benchmark #11565
Workflow file for this run
This file contains 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
| name: Compare VM performance to base branch | |
| on: | |
| pull_request: | |
| paths: | |
| - 'core/**' | |
| jobs: | |
| vm-benchmarks: | |
| name: Run VM benchmarks | |
| runs-on: [ matterlabs-ci-runner-highmem-long ] | |
| steps: | |
| - name: checkout base branch | |
| uses: actions/checkout@a5ac7e51b41094c92402da3b24376905380afc29 # v4 | |
| with: | |
| submodules: "recursive" | |
| fetch-depth: 0 | |
| ref: ${{ github.base_ref }} | |
| - name: fetch PR branch | |
| run: | | |
| git remote add pr_repo ${{ github.event.pull_request.head.repo.clone_url }} | |
| git fetch pr_repo ${{ github.event.pull_request.head.ref }} | |
| - name: fetch merge-base SHA | |
| id: merge_base | |
| run: echo "sha=$(git merge-base HEAD FETCH_HEAD)" >> $GITHUB_OUTPUT | |
| - name: checkout divergence point | |
| run: git checkout ${{ steps.merge_base.outputs.sha }} --recurse-submodules | |
| - name: setup-env | |
| run: | | |
| touch .env | |
| echo ZKSYNC_HOME=$(pwd) >> $GITHUB_ENV | |
| echo $(pwd)/bin >> $GITHUB_PATH | |
| echo $(pwd)/zkstack_cli/zkstackup >> $GITHUB_PATH | |
| echo "SCCACHE_GCS_BUCKET=matterlabs-infra-sccache-storage" >> .env | |
| echo "SCCACHE_GCS_SERVICE_ACCOUNT=gha-ci-runners@matterlabs-infra.iam.gserviceaccount.com" >> .env | |
| echo "SCCACHE_GCS_RW_MODE=READ_WRITE" >> .env | |
| echo "RUSTC_WRAPPER=sccache" >> .env | |
| # Set the minimum reported instruction count difference to reduce noise | |
| echo "BENCHMARK_DIFF_THRESHOLD_PERCENT=2" >> .env | |
| - name: init | |
| run: | | |
| run_retried docker compose pull zk | |
| docker compose up -d zk | |
| - name: run benchmarks on base branch | |
| shell: bash | |
| run: | | |
| ci_run zkstackup -g --local | |
| ci_run zkstack dev contracts --system-contracts | |
| ci_run cargo bench --package vm-benchmark --bench instructions -- --verbose || echo "Instructions benchmark is missing" | |
| ci_run cargo run --package vm-benchmark --release --bin instruction_counts | tee base-opcodes | |
| - name: checkout PR | |
| run: | | |
| git checkout --force FETCH_HEAD --recurse-submodules | |
| - name: run benchmarks on PR | |
| shell: bash | |
| id: comparison | |
| run: | | |
| ci_run zkstackup -g --local | |
| ci_run zkstack dev contracts --system-contracts | |
| ci_run cargo bench --package vm-benchmark --bench instructions -- --verbose | |
| EOF=$(dd if=/dev/urandom bs=15 count=1 status=none | base64) | |
| echo "speedup<<$EOF" >> $GITHUB_OUTPUT | |
| # Output all lines starting from the "## ..." comparison header. Ignore the `grep` error if no lines were matched | |
| ci_run cargo bench --package vm-benchmark --bench instructions -- --print | grep -A 1000 --color=never -e '^## ' >> $GITHUB_OUTPUT || [ "$?" = 1 ] | |
| echo "$EOF" >> $GITHUB_OUTPUT | |
| echo "opcodes<<$EOF" >> $GITHUB_OUTPUT | |
| ci_run cargo run --package vm-benchmark --release --bin instruction_counts -- --diff base-opcodes | grep -A 1000 --color=never -e '^## ' >> $GITHUB_OUTPUT || [ "$?" = 1 ] | |
| echo "$EOF" >> $GITHUB_OUTPUT | |
| - name: Comment on PR | |
| uses: thollander/actions-comment-pull-request@fabd468d3a1a0b97feee5f6b9e499eab0dd903f6 # v2.5.0 | |
| if: steps.comparison.outputs.speedup != '' || steps.comparison.outputs.opcodes != '' | |
| with: | |
| message: | | |
| ${{ steps.comparison.outputs.speedup }} | |
| ${{ steps.comparison.outputs.opcodes }} | |
| comment_tag: vm-performance-changes | |
| mode: recreate | |
| create_if_not_exists: true | |
| - name: Remove PR comment | |
| uses: thollander/actions-comment-pull-request@fabd468d3a1a0b97feee5f6b9e499eab0dd903f6 # v2.5.0 | |
| if: steps.comparison.outputs.speedup == '' && steps.comparison.outputs.opcodes == '' | |
| with: | |
| comment_tag: vm-performance-changes | |
| message: 'No performance difference detected (anymore)' | |
| mode: delete |