Is there any seamless way available with best fp16 performance being achieved in NV V100/P100? E.g. I've a model and implementation being trained in fp32. The App works perfectly. Now, I'd like to explore the experience of fp16. Is there any simple way to enable this.
1 Answers
try this method, and I found inference with fp16 is faster on Pascal architecture GPU, can someone give an explanation?
Related