| Joule / Graph | Device (GPU) | Power / W | Graph / s | data source |
|---|---|---|---|---|
| 450 | NVIDIA RTX 4080 | 180 | 0.40 | nicehash.com |
| 262 | NVIDIA RTX 3070 | 170 | 0.65 | nicehash.com |
| 199 | NVIDIA RTX 3090 | 285 | 1.43 | nicehash.com |
| 264 | AMD RX 7900 XT 20GB | 335 | 1.27 | nicehash.com |
| 658 | AMD RX Vega 64 | 250 | 0.38 | nicehash.com |
| 172 | AMD RX 6800 XT 16GB | 150 | 0.87 | nicehash.com |
| 140 | NVIDIA 4070Ti | 140 | 1.00 | whattomine.com |
| 185 | NVIDIA 4090 | 250 | 1.35 | whattomine.com |
| 179 | NVIDIA 4080 | 250 | 1.40 | whattomine.com |
| 292 | NVIDIA 3090Ti | 380 | 1.30 | whattomine.com |
| 276 | NVIDIA 3090 | 290 | 1.05 | whattomine.com |
| 243 | NVIDIA 3080Ti | 280 | 1.15 | whattomine.com |
| 167 | AMD 6800 | 150 | 0.90 | whattomine.com |
| 650 | AMD Vega64 | 260 | 0.40 | whattomine.com |
With thanks to nicehash.com and whattomine.com where I collected the data today. On this data NVIDIA 4070Ti is now more efficent than AMD 6800. But like @Neo mentioned in Discussion: Bounty for new GPU solver with better efficacy than a G1 mini it is likely that optimized solvers would impact the results.