[llvm] [NVPTX][Docs] [NFC] Update docs on intrinsics (PR #133136)

[Alex MacLean via llvm-commits]

================
@@ -153,6 +153,68 @@ Example: 64-bit PTX for CUDA Driver API: ``nvptx64-nvidia-cuda``
 NVPTX Intrinsics
 ================
 
+Guidelines on NVPTX Intrinsic design
----------------
AlexMaclean wrote:

I'm not sure where exactly this does belong but NVPTXUsage.rst seems like not the right place. This document is for users of the NVPTX backend and provides a user facing explanation for what IR to use and how it is lowered. This section is only relevant to developers of the NVPTX backend. Maybe it would be better as a comment in IntrinsicsNVVM.td?

https://github.com/llvm/llvm-project/pull/133136