NagyDonat wrote: You're right, repeating the same measurement produced significantly different results:
| Project | before this PR | this PR | Ratio | | --- | --- | --- | --- | | memcached | 25.70 | 28.12 | 109.4% | | tmux | 74.18 | 74.30 | 100.2% | | curl | 219.55 | 217.06 | 98.9% | | twin | 52.27 | 47.75 | 91.4% | | vim | 167.13 | 163.41 | 97.8% | | openssl | 431.63 | 422.84 | 98.0% | | sqlite | 959.32 | 941.19 | 98.1% | | ffmpeg | 1353.04 | 1348.64 | 99.7% | | postgres | 436.02 | 425.13 | 97.5% | | tinyxml2 | 31.80 | 32.39 | 101.9% | | libwebm | 67.39 | 65.77 | 97.6% | | xerces | 201.98 | 225.78 | 111.8% | | bitcoin | 175.57 | 173.28 | 98.7% | | protobuf | 2585.72 | 2367.59 | 91.6% | | qtbase | 3988.95 | 3984.71 | 99.9% | | openrct2 | 1210.14 | 1175.08 | 97.1% | | total | 11980.39 | 11693.04 | 97.6% | At least now I know that our environment is noisy and I will need to repeat measurements multiple times to get usable results :sweat_smile: https://github.com/llvm/llvm-project/pull/139939 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
