[PATCH] D23477: [NVPTX] Use ldg for explicitly invariant loads.

[Artem Belevich via llvm-commits]

tra accepted this revision.
tra added a comment.
This revision is now accepted and ready to land.

You may want to add a change for to make sure explicit invariant loads work within kernels, too.


================
Comment at: llvm/test/CodeGen/NVPTX/ldg-invariant.ll:21-22
@@ +20,4 @@
+; CHECK-LABEL: @ld_not_global_addrspace
+define i32 @ld_not_global_addrspace(i32 addrspace(0)* %ptr) {
+; CHECK: ld.{{[a-z]}}32
+  %a = load i32, i32 addrspace(0)* %ptr
----------------
You may want to add a test case for invariant load from non-global space.


https://reviews.llvm.org/D23477