[PATCH] D130090: [PowerPC][NFC] Convert the MMA test cases to use opaque pointers.

Stefan Pintilie via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Jul 21 18:28:41 PDT 2022


stefanp added inline comments.


================
Comment at: llvm/test/CodeGen/PowerPC/mma-integer-based-outer-product.ll:70
 ; Function Attrs: nofree nounwind
-define dso_local void @test3(i8* nocapture readonly %vqp, i8* nocapture readnone %vpp, <16 x i8> %vc, i8* nocapture %resp) {
+define dso_local void @test3(ptr nocapture readonly %vqp, ptr nocapture readnone %vpp, <16 x i8> %vc, ptr nocapture %resp) {
 ; CHECK-LABEL: test3:
----------------
amyk wrote:
> If `%vpp` is not used, do we still need to keep them? (Here and in the functions below)
You are correct that this is not used.
I had noticed that when I originally edited the test and I was debating whether or not to fix it. 
There are other tests that were written around the same time that have a similar issue
(see `test/CodeGen/PowerPC/mma-outer-product.ll` as well) 

The reason I chose not to do this cleanup was because that was not really the point of this patch and because it would be a fairly significant change and so the much larger diff would distract from the rest of the changes.

I could however do this as a later patch. 


================
Comment at: llvm/test/CodeGen/PowerPC/mma-intrinsics.ll:616
   %1 = tail call <512 x i1> @llvm.ppc.mma.xvf32gerpp(<512 x i1> %0, <16 x i8> %vc, <16 x i8> %vc)
-  %call = tail call signext i32 bitcast (i32 ()* @testRedundantPrimeUnprimeF to i32 ()*)()
-  %add.ptr1 = getelementptr inbounds <512 x i1>, <512 x i1>* %dst, i64 1
-  store <512 x i1> %1, <512 x i1>* %add.ptr1, align 64
+  %call = tail call signext i32 bitcast (ptr @testRedundantPrimeUnprimeF to ptr)()
+  %add.ptr1 = getelementptr inbounds <512 x i1>, ptr %dst, i64 1
----------------
lei wrote:
> is this bitcast still needed?
No, it's not needed. I think I missed that on my fist pass through.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130090/new/

https://reviews.llvm.org/D130090



More information about the llvm-commits mailing list