[X86] Add FMA commuting test case for D75016

This test case shows extra moves due to not fully considering all
commuting opportunities.

Author: topperc

llvm/test/CodeGen/X86/fma-commute-loop.ll

@@ -0,0 +1,116 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=avx512f | FileCheck %s
+
+define void @eggs(<8 x double>* %arg, <8 x double>* %arg1, <8 x double>* %arg2, <8 x double>* %arg3, <8 x double>* %arg4, <8 x double>* %arg5, i64 %arg6, i64 %arg7, i64 %arg8, i64 %arg9, i64 %arg10, i64 %arg11, i64 %arg12, double* %arg13, double* %arg14) nounwind {
+; CHECK-LABEL: eggs:
+; CHECK:       ## %bb.0: ## %bb
+; CHECK-NEXT:    pushq %r15
+; CHECK-NEXT:    pushq %r14
+; CHECK-NEXT:    pushq %r13
+; CHECK-NEXT:    pushq %r12
+; CHECK-NEXT:    pushq %rbx
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r10
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r11
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r12
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r15
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r14
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %rbx
+; CHECK-NEXT:    leaq (%rbx,%r14,8), %r14
+; CHECK-NEXT:    leaq (%rbx,%r15,8), %r15
+; CHECK-NEXT:    vxorpd %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    xorl %ebx, %ebx
+; CHECK-NEXT:    movq {{[0-9]+}}(%rsp), %r13
+; CHECK-NEXT:    addq %r12, %r13
+; CHECK-NEXT:    addq {{[0-9]+}}(%rsp), %r12
+; CHECK-NEXT:    vxorpd %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vxorpd %xmm2, %xmm2, %xmm2
+; CHECK-NEXT:    vxorpd %xmm4, %xmm4, %xmm4
+; CHECK-NEXT:    vxorpd %xmm5, %xmm5, %xmm5
+; CHECK-NEXT:    vxorpd %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    .p2align 4, 0x90
+; CHECK-NEXT:  LBB0_1: ## %bb15
+; CHECK-NEXT:    ## =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vmovapd %zmm5, %zmm6
+; CHECK-NEXT:    vmovapd %zmm4, %zmm7
+; CHECK-NEXT:    vmovupd (%rax,%r11,8), %zmm4
+; CHECK-NEXT:    vmovupd (%rax,%r13,8), %zmm5
+; CHECK-NEXT:    vmovupd (%rax,%r12,8), %zmm8
+; CHECK-NEXT:    vbroadcastsd (%r15,%rbx,8), %zmm9
+; CHECK-NEXT:    vfmadd231pd {{.*#+}} zmm0 = (zmm4 * zmm9) + zmm0
+; CHECK-NEXT:    vfmadd231pd {{.*#+}} zmm1 = (zmm5 * zmm9) + zmm1
+; CHECK-NEXT:    vfmadd231pd {{.*#+}} zmm2 = (zmm8 * zmm9) + zmm2
+; CHECK-NEXT:    vbroadcastsd (%r14,%rbx,8), %zmm9
+; CHECK-NEXT:    vfmadd213pd {{.*#+}} zmm4 = (zmm9 * zmm4) + zmm7
+; CHECK-NEXT:    vfmadd213pd {{.*#+}} zmm5 = (zmm9 * zmm5) + zmm6
+; CHECK-NEXT:    vfmadd231pd {{.*#+}} zmm3 = (zmm8 * zmm9) + zmm3
+; CHECK-NEXT:    incq %rbx
+; CHECK-NEXT:    cmpq %rbx, %r10
+; CHECK-NEXT:    jne LBB0_1
+; CHECK-NEXT:  ## %bb.2: ## %bb51
+; CHECK-NEXT:    vmovapd %zmm0, (%rdi)
+; CHECK-NEXT:    vmovapd %zmm1, (%rsi)
+; CHECK-NEXT:    vmovapd %zmm2, (%rdx)
+; CHECK-NEXT:    vmovapd %zmm4, (%rcx)
+; CHECK-NEXT:    vmovapd %zmm5, (%r8)
+; CHECK-NEXT:    vmovapd %zmm3, (%r9)
+; CHECK-NEXT:    popq %rbx
+; CHECK-NEXT:    popq %r12
+; CHECK-NEXT:    popq %r13
+; CHECK-NEXT:    popq %r14
+; CHECK-NEXT:    popq %r15
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    retq
+bb:
+  br label %bb15
+
+bb15:                                             ; preds = %bb15, %bb
+  %tmp = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp38, %bb15 ]
+  %tmp16 = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp39, %bb15 ]
+  %tmp17 = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp40, %bb15 ]
+  %tmp18 = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp46, %bb15 ]
+  %tmp19 = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp47, %bb15 ]
+  %tmp20 = phi <8 x double> [ zeroinitializer, %bb ], [ %tmp48, %bb15 ]
+  %tmp21 = phi i64 [ 0, %bb ], [ %tmp49, %bb15 ]
+  %tmp22 = getelementptr inbounds double, double* %arg14, i64 %arg11
+  %tmp23 = bitcast double* %tmp22 to <8 x double>*
+  %tmp24 = load <8 x double>, <8 x double>* %tmp23, align 8
+  %tmp25 = add i64 %arg10, %arg6
+  %tmp26 = getelementptr inbounds double, double* %arg14, i64 %tmp25
+  %tmp27 = bitcast double* %tmp26 to <8 x double>*
+  %tmp28 = load <8 x double>, <8 x double>* %tmp27, align 8
+  %tmp29 = add i64 %arg10, %arg7
+  %tmp30 = getelementptr inbounds double, double* %arg14, i64 %tmp29
+  %tmp31 = bitcast double* %tmp30 to <8 x double>*
+  %tmp32 = load <8 x double>, <8 x double>* %tmp31, align 8
+  %tmp33 = add i64 %tmp21, %arg8
+  %tmp34 = getelementptr inbounds double, double* %arg13, i64 %tmp33
+  %tmp35 = load double, double* %tmp34, align 8
+  %tmp36 = insertelement <8 x double> undef, double %tmp35, i32 0
+  %tmp37 = shufflevector <8 x double> %tmp36, <8 x double> undef, <8 x i32> zeroinitializer
+  %tmp38 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp24, <8 x double> %tmp37, <8 x double> %tmp)
+  %tmp39 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp28, <8 x double> %tmp37, <8 x double> %tmp16)
+  %tmp40 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp32, <8 x double> %tmp37, <8 x double> %tmp17)
+  %tmp41 = add i64 %tmp21, %arg9
+  %tmp42 = getelementptr inbounds double, double* %arg13, i64 %tmp41
+  %tmp43 = load double, double* %tmp42, align 8
+  %tmp44 = insertelement <8 x double> undef, double %tmp43, i32 0
+  %tmp45 = shufflevector <8 x double> %tmp44, <8 x double> undef, <8 x i32> zeroinitializer
+  %tmp46 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp24, <8 x double> %tmp45, <8 x double> %tmp18)
+  %tmp47 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp28, <8 x double> %tmp45, <8 x double> %tmp19)
+  %tmp48 = call <8 x double> @llvm.fmuladd.v8f64(<8 x double> %tmp32, <8 x double> %tmp45, <8 x double> %tmp20)
+  %tmp49 = add nuw nsw i64 %tmp21, 1
+  %tmp50 = icmp eq i64 %tmp49, %arg12
+  br i1 %tmp50, label %bb51, label %bb15
+
+bb51:                                             ; preds = %bb15
+  store <8 x double> %tmp38, <8 x double>* %arg
+  store <8 x double> %tmp39, <8 x double>* %arg1
+  store <8 x double> %tmp40, <8 x double>* %arg2
+  store <8 x double> %tmp46, <8 x double>* %arg3
+  store <8 x double> %tmp47, <8 x double>* %arg4
+  store <8 x double> %tmp48, <8 x double>* %arg5
+  ret void
+}
+
+declare <8 x double> @llvm.fmuladd.v8f64(<8 x double>, <8 x double>, <8 x double>)