[Performance] Severe performance penalty with transformer model and DirectML about onnxruntime HOT 5 OPEN

@yuslepukhin

I could collect some more runtime measurements, using the publicly available python bindings (v1.18.1). The results are interesting: The DirectML bindings are the fastest, at 1.5s per image, CUDA is at least as slow. If I run the same model with pytorch+CUDA on windows, I get ~0.5s per image, which is the same runtime I get on linux on identical hardware. On linux, our C++ onnxruntime integration is faster than pytorch by ~20%. It seems that the model has some issue on windows in onnxruntime, not clearly related to the execution provider. As from my original message, the issue seems to come from some extra memcpy, but I don't really understand what is causing them.

from onnxruntime.

@smk2007

from onnxruntime.

I made some extra experiments with dynamic axes. I could confirm that removing any dynamic axes provides a small speed-up (few % with DirectML), but once again it does not fill the gap between Windows+DirectML and Linux+CUDA.

from onnxruntime.

Any reason not to compare Windows CUDA with Linux CUDA?

from onnxruntime.

Any reason not to compare Windows CUDA with Linux CUDA?

The reason is that I don't have a have a reliable way to build onnxruntime with CUDA on Windows, cudart/cudnn distribution for Windows has been sketchy at best in recent times (at least up to 11.7 which is what I currently use). So, I cannot do the measurements on the same infrastructure I use for the other cases. A hacky test however suggests that Windows+CUDA suffers from a similar performance penalty than Windows+DirectML.

from onnxruntime.