wang-jiahua commented on PR #10536: URL: https://github.com/apache/rocketmq/pull/10536#issuecomment-4766521234
## Server Benchmark Results (2026-06-22) **Environment**: 8C30G broker, Dragonwell JDK 21, 1KB msg, 128 queues, sync mode, 256 threads, 20s warmup + 30s collect ### Producer + Consumer (steady state, lines 3-5) | Test | Producer TPS | Producer RT | Consumer TPS | Consumer RT | GC | vs Baseline | |------|-------------|-------------|-------------|-------------|----|-------------| | Baseline | 186k | 1.38ms | 187k | 167ms | 204 | — | | #10536 | 198k | 1.29ms | 200k | 170ms | 100 | **Prod +6.5%, Con +7.0%, GC -51%** ✅ | Pull header FastCodesHeader encode/decode eliminates reflection on every Pull request. In producer-only tests, the unused Pull code path caused JIT type profile pollution (-6.5%). With consumer running, the Pull path is exercised and shows +6.5% producer TPS, +7.0% consumer TPS, and 51% GC reduction. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
