[AArch64] Enable post-RA MI scheduler.

[oota-llvm.git] / test / CodeGen / AArch64 / postra-mi-sched.ll

diff --git a/test/CodeGen/AArch64/postra-mi-sched.ll b/test/CodeGen/AArch64/postra-mi-sched.ll
new file mode 100644 (file)
index 0000000..5a40724
--- /dev/null
+++ b/test/CodeGen/AArch64/postra-mi-sched.ll
@@ -0,0 +1,31 @@
+; RUN: llc < %s -O3 -march=aarch64 -mcpu=cortex-a53 | FileCheck %s
+
+; With cortex-a53, each of fmul and fcvt have latency of 6 cycles.  After the
+; pre-RA MI scheduler, fmul, fcvt and fdiv will be consecutive.  The top-down
+; post-RA MI scheduler will clean this up.
+
+@d1 = common global double 0.000000e+00, align 8
+
+define i32 @test1(float %s2, float %s3, double %d, i32 %i2, i32 %i3) {
+entry:
+; CHECK-LABEL: @test1
+; CHECK: fmul
+; CHECK-NEXT: add
+; CHECK: fcvt
+; CHECK-NEXT: mul
+  %mul = fmul float %s2, %s3
+  %conv = fpext float %mul to double
+  %div = fdiv double %d, %conv
+  store double %div, double* @d1, align 8
+  %factor = shl i32 %i3, 1
+  %add1 = add i32 %i2, 4
+  %add2 = add i32 %add1, %factor
+  %add3 = add nsw i32 %add2, %i2
+  %add4 = add nsw i32 %add3, %add2
+  %mul5 = mul i32 %add3, %add3
+  %mul6 = mul i32 %mul5, %add4
+  %mul7 = shl i32 %add4, 1
+  %factor18 = mul i32 %mul7, %mul6
+  %add9 = add i32 %factor18, %mul6
+  ret i32 %add9
+}