By: Kaapo Seppänen <seppanenkaapo@gmail.com>
lengthen the GL16 evaluator-probe benchmarks to 0.75 seconds Since bench stopped timing model resolution and warm-up, each 0.2-second GL16 evaluator benchmark is a 0.18-second loop of pure evaluation. On GitHub runners its paired warm-runtime noise reached 5-6%, so unchanged Iterative and SummedFunctionMap runtimes failed the 15% noise screen. At 0.75 seconds a forced discovery-plus-confirmation comparison took 182 s on itphlies, within the 240-second per-comparison budget. Keep NixCI disabled while this is in progress.
Total
1s
| Time to Start | Worker time | Duration | Time to finish | Idle | |
|---|---|---|---|---|---|
| Config | 0s | 1s | 1s | 1s | 0s |
| Eval | - | - | - | - | - |
| Build | - | - | - | - | - |
| Suite | 0s | 1s | 1s | 1s | 0s |