diff --git a/ppdiffusers/deploy/README.md b/ppdiffusers/deploy/README.md new file mode 100644 index 000000000..201da994f --- /dev/null +++ b/ppdiffusers/deploy/README.md @@ -0,0 +1,65 @@ +# PPDiffusers推理部署 + +PPDiffusers基于Paddle Inference,提供了以下重点扩散模型的部署方案: +- ControlNet +- IP-Adapter-SD15 +- IP-Adapter-SDXL +- SD15 +- SDXL + + +# V100性能数据 +|模型|Paddle Deploy TensorRT / ips|Torch Dynamic / ips| +|-|-|-| +|IP-Adapter-SD15 text2img|18.30|18.18| +|IP-Adapter-SD15 img2img|18.11|17.87| +|IP-Adapter-SD15 inpaint|17.93|17.44| +|IP-Adapter-SDXL text2img|12.01|11.47| +|IP-Adapter-SDXL img2img|12.00|10.95| +|IP-Adapter-SDXL inpaint|11.67|10.79| +|SD15 text2img|19.68|18.27| +|SD15 img2img|19.68|17.90| +|SD15 inpaint|19.44|17.56| +|SDXL text2img|13.91|11.50| +|SDXL img2img|13.86|11.60| +|SDXL inpaint|13.45|11.28| + + + +> Note: +> 测试环境或配置为Paddle 3.0 beta版本,V100 32G单卡,FP16。 +推理参数为Image Width = 512, Image Height = 512, Num Inference Steps = 50。 + +# A100性能数据 +|模型|Paddle Deploy TensorRT / ips|Torch Dynamic / ips| +|-|-|-| +|IP-Adapter-SD15 text2img|38.52|| +|IP-Adapter-SD15 img2img|37.91|| +|IP-Adapter-SD15 inpaint|37.80|| +|IP-Adapter-SDXL text2img|22.88|| +|IP-Adapter-SDXL img2img|22.79|| +|IP-Adapter-SDXL inpaint|22.30|| +|SD15 text2img|47.22|| +|SD15 img2img|46.59|| +|SD15 inpaint|46.05|| +|SDXL text2img|31.98|12.91| +|SDXL img2img|31.80|9.43| +|SDXL inpaint|30.58|11.29| + + + +> Note: 测试环境或配置为Paddle 3.0 beta版本,A100 80G单卡,FP16。 +推理参数为Image Width = 512, Image Height = 512, Num Inference Steps = 50。 + + + + \ No newline at end of file