[flang-commits] [clang] [flang] [llvm] clang: Start using new amdgpu subarch triples (PR #206482)

[Joseph Huber via flang-commits]

================
@@ -915,7 +915,9 @@ class ToolChain {
       return;
     }
 
-    if (TT.isAMDGPU()) {
+    if (TT.isAMDGCN()) {
----------------
jhuber6 wrote:

`amdgcn` is now distinct from `amdgpu`? It used to be that `amdgcn` was everything that wasn't r600, what is the model now?

https://github.com/llvm/llvm-project/pull/206482