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× | 24 / 57 | 23 / 56 | 2 / 6 |
| 2026-08-22 | 0.37× | 32 / 85 | 30 / 84 | 2 / 6 |
| 2026-08-23 | 0.37× | 86 / 153 | 79 / 148 | 2 / 6 |
| 2026-08-24 | 0.51× | 18 / 34 | 13 / 33 | 2 / 5 |
| 2026-08-26 | 0.26× | 54 / 161 | 35 / 153 | 2 / 6 |
| 2026-08-27 | 0.43× | 38 / 129 | 32 / 128 | 2 / 6 |
| 2026-08-29 | 0.40× | 75 / 43 | 64 / 43 | 2 / 6 |
| 2026-08-31 | 0.54× | 29 / 20 | 26 / 18 | 2 / 2 |
| 2026-09-03 | 0.90× | 96 / 39 | 95 / 39 | 2 / 5 |
The measured pair page for these deployments: side-by-side cells. Downloads: ratio cells, value board, the rental prices these were costed at.