[llvm] AMDGPU: Migrate DebugInfo tests to new subarch triples (58) (PR #209818)

[Pierre van Houtryve via llvm-commits]

================
@@ -1,11 +1,11 @@
-; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji < %s | FileCheck %s
+; RUN: llc -mtriple=amdgpu8.03-amd-amdhsa < %s | FileCheck %s
 
 ; Make sure we do not crash during scheduling when DBG_VALUE is the first
 ; instruction in the basic block.
 
 ; LLVM IR generated with the following command and OpenCL source:
 ;
-; $clang -cl-std=CL2.0 -g -O2 -target amdgcn-amd-amdhsa -S -emit-llvm <path-to-file>
+; $clang -cl-std=CL2.0 -g -O2 -target amdgpu8.03-amd-amdhsa -S -emit-llvm <path-to-file>
----------------
Pierre-vh wrote:

This test was using a default/generic target I assume, is it intentional to move it to gfx803? Same for a few others below

https://github.com/llvm/llvm-project/pull/209818