diff --git a/llvm/test/Regression/CodeGen/PowerPC/vcmp-fold.ll b/llvm/test/Regression/CodeGen/PowerPC/vcmp-fold.ll new file mode 100644 index 000000000000..04698d7d20b9 --- /dev/null +++ b/llvm/test/Regression/CodeGen/PowerPC/vcmp-fold.ll @@ -0,0 +1,21 @@ +; This should fold the "vcmpbfp." and "vcmpbfp" instructions into a single +; "vcmpbfp.". +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep vcmpbfp | wc -l | grep 1 + +void %test(<4 x float>* %x, <4 x float>* %y, int* %P) { +entry: + %tmp = load <4 x float>* %x ; <<4 x float>> [#uses=1] + %tmp2 = load <4 x float>* %y ; <<4 x float>> [#uses=1] + %tmp = call int %llvm.ppc.altivec.vcmpbfp.p( int 1, <4 x float> %tmp, <4 x float> %tmp2 ) ; [#uses=1] + %tmp4 = load <4 x float>* %x ; <<4 x float>> [#uses=1] + %tmp6 = load <4 x float>* %y ; <<4 x float>> [#uses=1] + %tmp = call <4 x int> %llvm.ppc.altivec.vcmpbfp( <4 x float> %tmp4, <4 x float> %tmp6 ) ; <<4 x int>> [#uses=1] + %tmp7 = cast <4 x int> %tmp to <4 x float> ; <<4 x float>> [#uses=1] + store <4 x float> %tmp7, <4 x float>* %x + store int %tmp, int* %P + ret void +} + +declare int %llvm.ppc.altivec.vcmpbfp.p(int, <4 x float>, <4 x float>) + +declare <4 x int> %llvm.ppc.altivec.vcmpbfp(<4 x float>, <4 x float>)