-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
perf(vm): Fix VM performance regression on CI loadtest #2782
Conversation
50ff4cd
to
a8838eb
Compare
a8838eb
to
b42ca4a
Compare
e6beefe
to
7ce9332
Compare
No performance difference detected (anymore) |
I haven't managed to find any reason for performance regression in recent PRs, and after reverting them, it looks like the load test sometimes fails even when these changes are fully reverted. Thus, I've Not sure whether it's just wishful thinking, but the changes in the load test config seem to make the load test more stable (decreasing the number of accounts and concurrent API requests; increasing the number of in-flight txs and randomizing polling interval between polling tx status). |
🤖 I have created a release *beep* *boop* --- ## [24.24.0](core-v24.23.0...core-v24.24.0) (2024-09-05) ### Features * conditional cbt l1 updates ([#2748](#2748)) ([6d18061](6d18061)) * **eth-watch:** do not query events from earliest block ([#2810](#2810)) ([1da3f7e](1da3f7e)) * **genesis:** Validate genesis config against L1 ([#2786](#2786)) ([b2dd9a5](b2dd9a5)) * Integrate tracers and implement circuits tracer in vm2 ([#2653](#2653)) ([87b02e3](87b02e3)) * Move prover data to /home/popzxc/workspace/current/zksync-era/prover/data ([#2778](#2778)) ([62e4d46](62e4d46)) * Remove prover db from house keeper ([#2795](#2795)) ([85b7346](85b7346)) * **vm-runner:** Implement batch data prefetching ([#2724](#2724)) ([d01840d](d01840d)) * **vm:** Extract batch executor to separate crate ([#2702](#2702)) ([b82dfa4](b82dfa4)) * **vm:** Simplify VM interface ([#2760](#2760)) ([c3bde47](c3bde47)) * **zk_toolbox:** add multi-chain CI integration test ([#2594](#2594)) ([05c940e](05c940e)) ### Bug Fixes * **config:** Do not panic for observability config ([#2639](#2639)) ([1e768d4](1e768d4)) * **core:** Batched event processing support for Reth ([#2623](#2623)) ([958dfdc](958dfdc)) * return correct witness inputs ([#2770](#2770)) ([2516e2e](2516e2e)) * **tee-prover:** increase retries to reduce spurious alerts ([#2776](#2776)) ([4fdc806](4fdc806)) * **tee-prover:** mitigate panic on redeployments ([#2764](#2764)) ([178b386](178b386)) * **tee:** lowercase enum TEE types ([#2798](#2798)) ([0f2f9bd](0f2f9bd)) * **vm-runner:** Fix statement timeouts in VM playground ([#2772](#2772)) ([d3cd553](d3cd553)) ### Performance Improvements * **vm:** Fix VM performance regression on CI loadtest ([#2782](#2782)) ([bc0d7d5](bc0d7d5)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). --------- Co-authored-by: zksync-era-bot <[email protected]>
What ❔
Fixes VM performance regression on CI loadtest introduced in #2760.
Why ❔
Changes in the VM interface made the VM eagerly clone compressed bytecodes if compression hasn't failed. Compressed bytecodes aren't used during sandboxed VM execution in the API server (the sandbox only checks that compression is successful).For new VMs, bytecodes can be borrowed from the VM state, which is what this PR does using
Cow
.Checklist
zk fmt
andzk lint
.