> it wouldn’t use functions like these and the generated comparable code would be on-pare performance wise Perhaps if XLA generated all functions from scratch,

> it wouldn’t use functions like these and the generated comparable code would be on-pare performance wise Perhaps if XLA generated all functions from scratch, this would be more compelling. But XLA relies very heavily on pattern-matching to common library functions (e.g. CuDNN), and these patterns will certainly work better on Nvidia GPUs than AMD GPUs. In this way, I actually think explicitly calling the common library functions is actually much more transparent.