A paid production workload on the reference fleet, measured on GeForce RTX 4080 and GeForce RTX 5090 on the same days. One workload, its own cells — never averaged with any other workload, because cross-workload aggregation is the substitution error this project exists to expose.
| DAY | GOODPUT GeForce RTX 4080 / GeForce RTX 5090 | ATTEMPTS | COMPLETIONS | CARDS |
|---|---|---|---|---|
| 2026-08-21 | 0.33× | 256 / 762 | 220 / 723 | 2 / 6 |
| 2026-08-22 | 0.34× | 244 / 813 | 215 / 777 | 2 / 6 |
| 2026-08-23 | 0.42× | 165 / 718 | 146 / 684 | 2 / 6 |
| 2026-08-24 | 0.56× | 41 / 123 | 39 / 113 | 2 / 6 |
| 2026-08-26 | 0.44× | 370 / 733 | 340 / 697 | 2 / 6 |
| 2026-08-27 | 0.38× | 368 / 663 | 325 / 630 | 2 / 6 |
| 2026-08-29 | 0.38× | 377 / 203 | 323 / 190 | 2 / 6 |
| 2026-08-31 | 0.37× | 195 / 21 | 176 / 21 | 2 / 2 |
| 2026-09-02 | 0.79× | 38 / 11 | 30 / 8 | 2 / 4 |
| 2026-09-03 | 0.81× | 932 / 220 | 855 / 205 | 2 / 5 |
| 2026-09-04 | 0.58× | 57 / 24 | 48 / 24 | 2 / 3 |
The measured pair page for these deployments: side-by-side cells. Downloads: ratio cells, value board, the rental prices these were costed at.