ipiszy

For now it's for single GPU inference only.

RTX 3080-10GB should work. You could check https://github.com/facebookincubator/AITemplate/tree/main/ex..., and https://www.reddit.com/r/StableDiffusion/comments/xv7m89/met....

Yes this is correct. batch 16 7.9s / 25 steps, per image 0.49s: it generates 16 images for each prompt within 7.9s, so it's 0.49s per image.

AITemplate only supports fp16 data types with fp16 or fp32 accumulation right now. We are working on supporting more data types and quantization.

We don't have an official comparison between AITemplate and tvm / onnx for now, but we do have perf numbers like https://github.com/facebookincubator/AITemplate/tree/main/ex..., https://github.com/facebookincubator/AITemplate/tree/main/ex.... Feel free to run these examples on other frameworks and compare perf.

We have a bunch of unittests and E2E tests to compare numeric numbers between AITemplate and PyTorch eager.

You could check "AITemplate optimizations" section in the blog (https://ai.facebook.com/blog/gpu-inference-engine-nvidia-amd...), and https://github.com/facebookincubator/AITemplate#more-about-a.... The basic idea is to do aggressive kernel fusions.

As @haolu7 mentioned, you could take a pre-trained model and use AITemplate to do model inference. All you need to do is to re-write the model using AITemplate frontend and map PyTorch params to AITemplate params. Besides, AITemplate has a limited operator coverage compared to mature frameworks like PyTorch so you may need to implement your own kernels if necessary (though it already supports Bert, VIT, StableDiffusion, ResNet, Detectron, and general recommendation models).

tl;dr:

Meta is open sourcing AITemplate, an inference engine for both Nvidia and AMD GPUs. Code: https://github.com/facebookincubator/AITemplate.

AITemplate delivers much better perf (1.9x ~ 12.8x) compared to PyTorch eager on SOTA models, including Bert, ResNet, VIT and StableDiffusion.

AITemplate also delivers high perf numbers using AMD GPUs (MI-250). With AITemplate, MI-250 achieves 80% ~ 96% A100 perf on various ResNet / Bert / VIT models.

AITemplate uses sophisticated fusion techniques to optimize perf, including vertical, horizontal, and memory fusions.

btw, I'm one of the authors of AITemplate, happy to answer any questions.

I like HackerNode more because:

1. It contains more columns than only "FrontPage", such as "Jobs" and "Comments"; 2. The UI seems more concise. Although there are not indent between comments, the text is much clearer.

However, neither of these apps can be used to add comments. So I can only come back to my computer and add this comment on the web page.

...............................