量化精度选择 FP8和INT8？

[imoldpan]

理论算力注意

4090算力

参考

• https://developer.nvidia.com/blog/tensorrt-accelerates-stable-diffusion-nearly-2x-faster-with-8-bit-post-training-quantization/
• https://www.reddit.com/r/StableDiffusion/comments/1baeo5h/nvidia_tensorrt_int8_fp8_quantization/
• https://zhuanlan.zhihu.com/p/574825662