================
@@ -683,6 +686,17 @@ def CreateHandle : DXILOp<57, createHandle> {
let stages = [Stages<DXIL1_0, [all_stages]>];
}
+def BufferLoad : DXILOp<68, bufferLoad> {
+ let Doc = "reads from a TypedBuffer";
+ // Handle, Coord0, Coord1
+ let arguments = [HandleTy, Int32Ty, Int32Ty];
+ let result = OverloadTy;
+ let overloads =
+ [Overloads<DXIL1_0,
+ [ResRetHalfTy, ResRetFloatTy, ResRetInt16Ty, ResRetInt32Ty]>];
----------------
llvm-beanz wrote:
The 16-bit overloads were always valid in DXIL 1.0, but they didn't actually
mean 16-bit types, they meant the min16{float|int|uint} types. This is one of
the things that's really wonky about DXIL defining interpretations of LLVM IR
that conflicted with LLVM's core definition.
I think the code here is accurate to what we need for that.
https://github.com/llvm/llvm-project/pull/104252
_______________________________________________
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
