[clang] [llvm] [NVPTX] Improve NVVMReflect Efficiency (PR #134416)

[Artem Belevich via cfe-commits]

================
@@ -1,26 +1,53 @@
-; Verify that when passing in command-line options to NVVMReflect, that reflect calls are replaced with
-; the appropriate command line values.
+; Test the NVVM reflect pass functionality: verifying that reflect calls are replaced with 
+; appropriate values based on command-line options. Verify that we can handle custom reflect arguments
+; that aren't __CUDA_ARCH or __CUDA_FTZ. If that argument is given a value on the command-line, the reflect call should be replaced with that value.
+; Otherwise, the reflect call should be replaced with 0.
+
+; RUN: opt -passes=nvvm-reflect -mtriple=nvptx-nvidia-cuda -nvvm-reflect-add __CUDA_FTZ=1 -nvvm-reflect-add __CUDA_ARCH=350 %s -S | FileCheck %s --check-prefix=CHECK-FTZ1 --check-prefix=CHECK-ARCH350 --check-prefix=CHECK-CUSTOM-ABSENT
+; RUN: opt -passes=nvvm-reflect -mtriple=nvptx-nvidia-cuda -nvvm-reflect-add __CUDA_FTZ=0 -nvvm-reflect-add __CUDA_ARCH=520 %s -S | FileCheck %s --check-prefix=CHECK-FTZ0 --check-prefix=CHECK-ARCH520 --check-prefix=CHECK-CUSTOM-ABSENT
----------------
Artem-B wrote:

Long run lines should be split into separate lines:

```
RUN: opt ..... \
RUN:   | FileCheck .....
```

https://github.com/llvm/llvm-project/pull/134416