2 ms·They are not optimized for inference vs RTX GPUs.by lostmsu 2y agoThey are not optimized for inference vs RTX GPUs.