[llvm] AMDGPU: Add SubArchSpelling override to the TargetParser TableGen (PR #218500)

[Chinmay Deshpande via llvm-commits]

================
@@ -108,6 +108,10 @@ enum FeatureError : uint32_t {
 };
 
 LLVM_ABI StringRef getArchFamilyNameAMDGCN(GPUKind AK);
+
+/// The canonical GPU name for a variant name.
+LLVM_ABI StringRef getBaseArchNameAMDGCN(GPUKind AK);
----------------
chinmaydd wrote:

Would appreciate some unit tests, but not a blocker 

https://github.com/llvm/llvm-project/pull/218500