[clang] [WIP][AMDGPU] Support for type inferring image load/store builtins for AMDGPU (PR #140210)

Shilei Tian via cfe-commits cfe-commits at lists.llvm.org
Fri May 16 08:56:48 PDT 2025


================
@@ -683,6 +683,30 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
 
     return Builder.CreateInsertElement(I0, A, 1);
   }
+  case AMDGPU::BI__builtin_amdgcn_image_load_2d_f32_i32: {
+    llvm::Type *RetTy = llvm::Type::getFloatTy(Builder.getContext());
----------------
shiltian wrote:

I don't think you would need special handling of this builtin. It can be mapped to the intrinsic directly.

https://github.com/llvm/llvm-project/pull/140210


More information about the cfe-commits mailing list