Increase number of runs for all performance benchmarks #1315
Labels
Impact: Medium
Performance
Issues related to the performance
Tech Debt
Tests
Related to tests or testing environment
After increasing the number of runs for basic performance benchmarks (currently 100 runs, #1303), I see much less noise in their results (see screenshot). The benchmarks will be much more useful to us if they were run less often but with greater accuracy (I'd make all benchmarks run 100 times).
Example performance report: first 4 benchmarks run 100 times, last 5 run only once:
The text was updated successfully, but these errors were encountered: