Repository navigation
Website: expose the concurrent QPS metric; fix false successes in the QPS test - #2492
Merged
Merged
Conversation
… QPS test Add the "Concurrent QPS" metric to the main page and a row in the detailed comparison. Only results with zero errors are shown, and only for systems without failed queries in the main run. The concurrent test counted a query as successful by the exit code of ./query alone, while the main run also requires the timing line. Jd's ijconsole exits with zero on errors, so its failed queries were counted as successful, giving 15..155 QPS instead of about 0.04. Use the same criterion in both, make jd/query exit with non-zero on errors, skip the concurrent test for Jd (an embedded engine), and remove the QPS from its results. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
alexey-milovidov
requested a deployment
to
benchmark-approval
October 9, 2026 20:25 — with 
GitHub Actions
Waiting
GitHub Actions
Waiting
This branch is waiting to be deployed
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This was suggested by @toschmidt during the conversation at Open House Munich.
Adds the Concurrent QPS metric (10 connections, see CHANGELOG 2026-05-11) to the main page: a new metric selector and a row in the detailed comparison. Absolute QPS is shown along with the relative value (best QPS / system QPS, lower is better, as for the other metrics). It is not included in the Combined score.
Only results with zero errors are shown. A result is also hidden if the system failed some queries in the main run: the concurrent test cycles through all queries, so zero errors there means the errors went undetected.
Fix in the concurrent test
The concurrent test counted a query as successful by the exit code of
./queryalone, while the main run also requires the timing line. Jd'sijconsoleexits with zero on errors, so its failed queries counted as fast successes: 15–155 QPS reported vs. ~0.04 implied by its sequential timings.lib/benchmark-common.sh: both paths use the same success criterion (bench_extract_timing).jd/query: exit non-zero when no timing was produced;query.ijsprints the timing in fixed-point (J writes1e_5otherwise).jd/benchmark.sh: skip the concurrent test by default, as for other embedded engines (concurrent_qps measures swap-thrashing, not throughput, for single-process engines (introduced in #860) #946).jd/results/20260721/*.json: remove the QPS fields.data.generated.jswill be rebuilt by the bot.🤖 Generated with Claude Code