[CUDA] Bump the latest supported CUDA version to 11.4.

[tra]

Summary

This should reduce the amount of noise issued by clang for the recent-ish CUDA versions.

Clang still does not support all the features offered by NVCC, but is expected
to handle CUDA headers and produce binaries for all GPUs supported by NVCC.

[Hahnfeld]

LG, should land after the other revisions are in.

[tra]

Prepend space to version string.

[tra]

Undo unintentional change.

[tra]

Updated release notes.

clang/docs/ReleaseNotes.rst

 C++20 Feature Support
 ^^^^^^^^^^^^^^^^^^^^^
 ...
 
 C++2b Feature Support
 ^^^^^^^^^^^^^^^^^^^^^
 ...
 
+CUDA Language Changes in Clang
+---------------------------
+
+- Clang now supports CUDA versions up to 11.4.
+- Default GPU architecture has been changed from sm_20 to sm_35.
+
 Objective-C Language Changes in Clang
 -------------------------------------
 
 OpenCL C Language Changes in Clang
 ----------------------------------
 
 ...
 

clang/include/clang/Basic/Cuda.h

enum class CudaVersion {
   CUDA_100,
   CUDA_101,
   CUDA_102,
   CUDA_110,
   CUDA_111,
   CUDA_112,
   CUDA_113,
   CUDA_114,
-  FULLY_SUPPORTED = CUDA_101,
+  FULLY_SUPPORTED = CUDA_114,
   PARTIALLY_SUPPORTED =
       CUDA_114, // Partially supported. Proceed with a warning.
   NEW = 10000,  // Too new. Issue a warning, but allow using it.
 };
 const char *CudaVersionToString(CudaVersion V);
 // Input is "Major.Minor"
 CudaVersion CudaStringToVersion(const llvm::Twine &S);