[PATCH] D158883: [Matrix] Try to emit fmuladd for both vector and matrix types

[Ulrich Weigand via Phabricator via cfe-commits]

uweigand added a comment.

The newly added test cases in ffp-model.c fail on SystemZ, making CI red:
https://lab.llvm.org/buildbot/#/builders/94/builds/16280

The root cause seems to be that by default, the SystemZ back-end targets a machine without SIMD support, and therefore vector return types are passed via implicit reference according to the ABI:

  /home/uweigand/sandbox/buildbot/clang-s390x-linux/llvm/clang/test/CodeGen/ffp-model.c:121:12: error: CHECK: expected string not found in input
   // CHECK: define{{.*}} <4 x float> @my_m22_muladd
             ^
  <stdin>:62:28: note: scanning from here
   %4 = fadd fast <2 x float> %2, %3
                             ^
  <stdin>:67:1: note: possible intended match here
  define dso_local void @my_m22_muladd(ptr noalias sret([4 x float]) align 4 %agg.result, ptr noundef %0, float noundef nofpclass(nan inf) %y, ptr noundef %1) #0 {
  ^


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D158883/new/

https://reviews.llvm.org/D158883