CodSpeedHQ/codspeed-jvm - CodSpeed

codspeed-jvm

Blog Docs Changelog

Performance History

Latest Results

fix(agents): remove stale Justfile profiling section

fix/readme-and-agents-cleanup

2 days ago

feat(jmh-fork): add CodSpeed mode to capture raw ops and duration per iteration Add Mode.CodSpeed, an internal benchmark mode that runs identically to AverageTime but preserves the raw (ops, durationNs) pair per iteration in a new CodSpeedResult class. This eliminates the lossy floating-point roundtrip that the previous approach required (score = time/ops, then back-calculate time = score * ops). When CodSpeed instrumentation is detected, the Runner automatically overrides all declared benchmark modes to CodSpeed and deduplicates by benchmark name so multi-mode declarations run exactly once. CodSpeedResultCollector is simplified to only handle Mode.CodSpeed, reading raw values directly from CodSpeedResult with no back-calculation.

main

2 days ago

feat(jmh-fork): add CodSpeed mode to capture raw ops and duration per iteration Add Mode.CodSpeed, an internal benchmark mode that runs identically to AverageTime but preserves the raw (ops, durationNs) pair per iteration in a new CodSpeedResult class. This eliminates the lossy floating-point roundtrip that the previous approach required (score = time/ops, then back-calculate time = score * ops). When CodSpeed instrumentation is detected, the Runner automatically overrides all declared benchmark modes to CodSpeed and deduplicates by benchmark name so multi-mode declarations run exactly once. CodSpeedResultCollector is simplified to only handle Mode.CodSpeed, reading raw values directly from CodSpeedResult with no back-calculation.

cod-2446-normalize-benchmarkmode-to-sampletime

2 days ago

fix: handle git worktrees in findGitRoot by checking for .git file or directory

main

2 days ago

feat(jmh-fork): add CodSpeed mode to capture raw ops and duration per iteration Add Mode.CodSpeed, an internal benchmark mode that runs identically to AverageTime but preserves the raw (ops, durationNs) pair per iteration in a new CodSpeedResult class. This eliminates the lossy floating-point roundtrip that the previous approach required (score = time/ops, then back-calculate time = score * ops). When CodSpeed instrumentation is detected, the Runner automatically overrides all declared benchmark modes to CodSpeed and deduplicates by benchmark name so multi-mode declarations run exactly once. CodSpeedResultCollector is simplified to only handle Mode.CodSpeed, reading raw values directly from CodSpeedResult with no back-calculation.

cod-2446-normalize-benchmarkmode-to-sampletime

2 days ago

ci: skip upload for non-temurin distribution benchmarks

main

2 days ago

fix: handle git worktrees in findGitRoot by checking for .git file or directory

cod-2447-remove-fork0-limitation-from-jmh

7 days ago

fixup! feat: normalize benchmark mode to AverageTime under CodSpeed

cod-2446-normalize-benchmarkmode-to-sampletime

8 days ago

Latest Branches

-19%

docs: cleanup README and AGENTS.md#6

2 days ago

76a2b88

fix/readme-and-agents-cleanup

+39%

feat: normalize benchmark mode to AverageTime under CodSpeed#5

2 days ago

8cb09f5

cod-2446-normalize-benchmarkmode-to-sampletime

-42%

fix: remove fork(0) limitation from JMH runner#4

7 days ago

4b7a767

cod-2447-remove-fork0-limitation-from-jmh

© 2026 CodSpeed Technology

Home Terms Privacy Docs