Text-to-Image Generation Inference Efficiency on SD inference efficiency 1.4
3.11Inference Time (s)SD 1.4
Evaluation Results
| Method | Links | |
|---|---|---|
| SD 1.4GPU=NVIDIA RTX 4090, batch size=1, image resolution=512×512, steps=50 DDIM steps2026.07 | 3.11 | |
| AEGISGPU=NVIDIA RTX 4090, batch size=1, image resolution=512×512, steps=50 DDIM steps2026.07 | 3.99 | |
| SLDGPU=NVIDIA RTX 4090, batch size=1, image resolution=512×512, steps=50 DDIM steps2026.07 | 4.39 | |
| SAFREEGPU=NVIDIA RTX 4090, batch size=1, image resolution=512×512, steps=50 DDIM steps2026.07 | 4.43 | |
| SAeUronGPU=NVIDIA RTX 4090, batch size=1, image resolution=512×512, steps=50 DDIM steps2026.07 | 11.6 |