Text-to-Image Generation on FLUX.1 1024x1024 resolution (dev)
1.095ImageRewardFLUX
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| FLUXSampling Steps=50, Latency (s)=25.1, Speedup (s)=1.00×, TFLOPs=2991.01, Resolution=1024x10242026.05 | 1.095 | 0.707 | 6.75 | 0.634 | 0.698 | |
| Spectral Progressive DiffusionSampling Steps=10, Spatial acceleration factor (S)=2, Accel.=S, Latency (s)=4.35, Speedup (s)=5.77×, TFLOPs=500.34, Resolution=1024x10242026.05 | 1.059 | 0.696 | 6.69 | 0.624 | 0.655 | |
| Spectral Progressive DiffusionSampling Steps=50, Spatial acceleration factor (S)=2, Accel.=S, Latency (s)=15.16, Speedup (s)=1.66×, TFLOPs=1755.22, Resolution=1024x10242026.05 | 1.049 | 0.719 | 6.43 | 0.617 | 0.654 | |
| Spectral Progressive DiffusionSampling Steps=10, Spatial acceleration factor (S)=3, Accel.=S, Latency (s)=4.12, Speedup (s)=6.09×, TFLOPs=469.15, Resolution=1024x10242026.05 | 1.042 | 0.701 | 6.53 | 0.623 | 0.637 | |
| Spectral Progressive DiffusionSampling Steps=7, Spatial acceleration factor (S)=2, Accel.=S, Latency (s)=3.70, Speedup (s)=6.78×, TFLOPs=427.03, Resolution=1024x10242026.05 | 1.039 | 0.689 | 6.78 | 0.62 | 0.667 | |
| RALUSampling Steps=50, Accel.=S, Latency (s)=15.85, Speedup (s)=1.58×, TFLOPs=1749.94, Resolution=1024x10242026.05 | 1.028 | 0.712 | 6.07 | 0.613 | 0.648 | |
| RALUSampling Steps=10, Accel.=S, Latency (s)=5.04, Speedup (s)=4.98×, TFLOPs=540.47, Resolution=1024x10242026.05 | 1.022 | 0.7 | 6.43 | 0.626 | 0.652 | |
| Spectral Progressive DiffusionSampling Steps=50, Spatial acceleration factor (S)=3, Accel.=S, Latency (s)=14.54, Speedup (s)=1.73×, TFLOPs=1672.04, Resolution=1024x10242026.05 | 1.015 | 0.711 | 6.33 | 0.593 | 0.64 | |
| Spectral Progressive DiffusionSampling Steps=7, Spatial acceleration factor (S)=3, Accel.=S, Latency (s)=3.54, Speedup (s)=7.09×, TFLOPs=406.24, Resolution=1024x10242026.05 | 1.015 | 0.694 | 5.99 | 0.627 | 0.637 | |
| RALUSampling Steps=7, Accel.=S, Latency (s)=3.75, Speedup (s)=6.69×, TFLOPs=426.01, Resolution=1024x10242026.05 | 0.999 | 0.681 | 6.87 | 0.633 | 0.682 | |
| FLUXSampling Steps=10, Accel.=T, Latency (s)=5.18, Speedup (s)=4.84×, TFLOPs=610.02, Resolution=1024x10242026.05 | 0.981 | 0.679 | 6.93 | 0.618 | 0.647 | |
| TaylorSeerSampling Steps=10, Accel.=T, Latency (s)=9.34, Speedup (s)=2.69×, TFLOPs=556.72, Resolution=1024x10242026.05 | 0.972 | 0.684 | 6.77 | 0.594 | 0.619 | |
| TeaCacheSampling Steps=10, Accel.=T, Latency (s)=5.23, Speedup (s)=4.80×, TFLOPs=610.59, Resolution=1024x10242026.05 | 0.944 | 0.665 | 7.92 | 0.62 | 0.647 | |
| FLUXSampling Steps=7, Accel.=T, Latency (s)=3.79, Speedup (s)=6.62×, TFLOPs=431.45, Resolution=1024x10242026.05 | 0.92 | 0.66 | 8.25 | 0.594 | 0.583 | |
| BottleneckSampling Steps=10, Accel.=S, Latency (s)=5.37, Speedup (s)=4.67×, TFLOPs=571.23, Resolution=1024x10242026.05 | 0.889 | 0.661 | 9.16 | 0.62 | 0.687 | |
| BottleneckSampling Steps=7, Accel.=S, Latency (s)=3.78, Speedup (s)=6.64×, TFLOPs=431.52, Resolution=1024x10242026.05 | 0.792 | 0.631 | 8.71 | 0.605 | 0.672 | |
| TeaCacheSampling Steps=7, Accel.=T, Latency (s)=4.21, Speedup (s)=5.96×, TFLOPs=431.83, Resolution=1024x10242026.05 | 0.733 | 0.623 | 13.7 | 0.599 | 0.594 | |
| TaylorSeerSampling Steps=7, Accel.=T, Latency (s)=7.00, Speedup (s)=3.59×, TFLOPs=431.74, Resolution=1024x10242026.05 | 0.66 | 0.646 | 9.43 | 0.514 | 0.446 | |
| Δ-DiTSampling Steps=10, Accel.=T, Latency (s)=7.42, Speedup (s)=3.38×, TFLOPs=772.10, Resolution=1024x10242026.05 | 0.102 | 0.487 | 9.6 | 0.306 | 0.397 | |
| ToCaSampling Steps=10, Accel.=T, Latency (s)=15.5, Speedup (s)=1.62×, TFLOPs=601.12, Resolution=1024x10242026.05 | -1.827 | 0.253 | 10.6 | 0.259 | 0.137 |