[PATCH] D60279: [CUDA] Implemented _[bi]mma* builtins.

[Tim Shen via Phabricator via cfe-commits]

timshen added inline comments.


================
Comment at: clang/lib/CodeGen/CGBuiltin.cpp:12884
+// Helper classes for mapping MMA builtins to particular LLVM intrinsic variant.
+class NVPTXMmaLdstInfo {
+public:
----------------
How about having a simple struct and a function?
```
struct NvptxMmaLdstInfo {
  unsigned NumResults;
  unsigned IID_col;
  unsigned IID_row;
};

NvptxMmaLdstInfo getNvptxMmaLdstInfo(unsigned BuiltinID) { ... }
```

I don't see the need for classes here.


================
Comment at: clang/lib/CodeGen/CGBuiltin.cpp:13020
+
+class NVPTXMmaInfo {
+private:
----------------
ditto (struct + function)?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D60279/new/

https://reviews.llvm.org/D60279