divijvaidya opened a new pull request, #13307: URL: https://github.com/apache/kafka/pull/13307
# Motivation The current implementation of the benchmark may produce different set of integers across benchmarks and hence, may not provide apples to apples comparison. # Changes 1. With this change, we initialize random number generator with a fixed seed at the start of a benchmark. This ensures that for each benchmark, random number generator produces the same sequence of random values. Hence, the input data across benchmarks would be consistent providing a reliable apples to apples comparison. 2. We ensure that a new set of random numbers are generated per iteration, so that the bechmark calculation is performed over a diverse range of values Sample result of a benchmark run: ``` Benchmark Mode Cnt Score Error Units ByteUtilsBenchmark.testSizeOfUnsignedVarint thrpt 30 1302.193 ± 0.396 ops/s ByteUtilsBenchmark.testSizeOfUnsignedVarintSimple thrpt 30 328.678 ± 0.269 ops/s ByteUtilsBenchmark.testSizeOfVarlong thrpt 30 880.113 ± 0.676 ops/s ByteUtilsBenchmark.testSizeOfVarlongSimple thrpt 30 109.592 ± 0.071 ops/s JMH benchmarks done ``` -- 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: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org