Use vAny type to get rid of Neon intrinsics that differed only in whether
the overloaded vector types allowed floating-point or integer vector elements.
Most of these operations actually depend on the element type, so bitcasting
was not an option.
If you include the vpadd intrinsics that I updated earlier, this gets rid
of 20 intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78646 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/ARM/vrecpe.ll b/test/CodeGen/ARM/vrecpe.ll
index 79cb595..622725b 100644
--- a/test/CodeGen/ARM/vrecpe.ll
+++ b/test/CodeGen/ARM/vrecpe.ll
@@ -16,18 +16,18 @@
define <2 x float> @vrecpef32(<2 x float>* %A) nounwind {
%tmp1 = load <2 x float>* %A
- %tmp2 = call <2 x float> @llvm.arm.neon.vrecpef.v2f32(<2 x float> %tmp1)
+ %tmp2 = call <2 x float> @llvm.arm.neon.vrecpe.v2f32(<2 x float> %tmp1)
ret <2 x float> %tmp2
}
define <4 x float> @vrecpeQf32(<4 x float>* %A) nounwind {
%tmp1 = load <4 x float>* %A
- %tmp2 = call <4 x float> @llvm.arm.neon.vrecpef.v4f32(<4 x float> %tmp1)
+ %tmp2 = call <4 x float> @llvm.arm.neon.vrecpe.v4f32(<4 x float> %tmp1)
ret <4 x float> %tmp2
}
declare <2 x i32> @llvm.arm.neon.vrecpe.v2i32(<2 x i32>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vrecpe.v4i32(<4 x i32>) nounwind readnone
-declare <2 x float> @llvm.arm.neon.vrecpef.v2f32(<2 x float>) nounwind readnone
-declare <4 x float> @llvm.arm.neon.vrecpef.v4f32(<4 x float>) nounwind readnone
+declare <2 x float> @llvm.arm.neon.vrecpe.v2f32(<2 x float>) nounwind readnone
+declare <4 x float> @llvm.arm.neon.vrecpe.v4f32(<4 x float>) nounwind readnone