1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE
3 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX
5 define { i64, i64 } @PR46189(double %0, double %1, double %2, double %3, double %4) {
8 ; SSE-NEXT: movapd %xmm0, %xmm5
9 ; SSE-NEXT: subsd %xmm2, %xmm5
10 ; SSE-NEXT: addsd %xmm2, %xmm0
11 ; SSE-NEXT: unpcklpd {{.*#+}} xmm5 = xmm5[0],xmm0[0]
12 ; SSE-NEXT: unpcklpd {{.*#+}} xmm3 = xmm3[0,0]
13 ; SSE-NEXT: divpd %xmm3, %xmm5
14 ; SSE-NEXT: cvttpd2dq %xmm5, %xmm0
15 ; SSE-NEXT: movapd %xmm1, %xmm3
16 ; SSE-NEXT: subsd %xmm2, %xmm3
17 ; SSE-NEXT: addsd %xmm2, %xmm1
18 ; SSE-NEXT: unpcklpd {{.*#+}} xmm3 = xmm3[0],xmm1[0]
19 ; SSE-NEXT: unpcklpd {{.*#+}} xmm4 = xmm4[0,0]
20 ; SSE-NEXT: divpd %xmm4, %xmm3
21 ; SSE-NEXT: cvttpd2dq %xmm3, %xmm1
22 ; SSE-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
23 ; SSE-NEXT: movq %xmm0, %rax
24 ; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
25 ; SSE-NEXT: movq %xmm0, %rdx
30 ; AVX-NEXT: vsubsd %xmm2, %xmm0, %xmm5
31 ; AVX-NEXT: vaddsd %xmm2, %xmm0, %xmm0
32 ; AVX-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm5[0],xmm0[0]
33 ; AVX-NEXT: vmovddup {{.*#+}} xmm3 = xmm3[0,0]
34 ; AVX-NEXT: vdivpd %xmm3, %xmm0, %xmm0
35 ; AVX-NEXT: vcvttpd2dq %xmm0, %xmm0
36 ; AVX-NEXT: vsubsd %xmm2, %xmm1, %xmm3
37 ; AVX-NEXT: vaddsd %xmm2, %xmm1, %xmm1
38 ; AVX-NEXT: vunpcklpd {{.*#+}} xmm1 = xmm3[0],xmm1[0]
39 ; AVX-NEXT: vmovddup {{.*#+}} xmm2 = xmm4[0,0]
40 ; AVX-NEXT: vdivpd %xmm2, %xmm1, %xmm1
41 ; AVX-NEXT: vcvttpd2dq %xmm1, %xmm1
42 ; AVX-NEXT: vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
43 ; AVX-NEXT: vmovq %xmm0, %rax
44 ; AVX-NEXT: vpextrq $1, %xmm0, %rdx
46 %6 = insertelement <2 x double> undef, double %0, i32 0
47 %7 = shufflevector <2 x double> %6, <2 x double> undef, <2 x i32> zeroinitializer
48 %8 = insertelement <2 x double> undef, double %2, i32 0
49 %9 = shufflevector <2 x double> %8, <2 x double> undef, <2 x i32> zeroinitializer
50 %10 = fsub <2 x double> %7, %9
51 %11 = fadd <2 x double> %7, %9
52 %12 = shufflevector <2 x double> %10, <2 x double> %11, <2 x i32> <i32 0, i32 3>
53 %13 = insertelement <2 x double> undef, double %3, i32 0
54 %14 = shufflevector <2 x double> %13, <2 x double> undef, <2 x i32> zeroinitializer
55 %15 = fdiv <2 x double> %12, %14
56 %16 = fptosi <2 x double> %15 to <2 x i32>
57 %17 = insertelement <2 x double> undef, double %1, i32 0
58 %18 = shufflevector <2 x double> %17, <2 x double> undef, <2 x i32> zeroinitializer
59 %19 = fsub <2 x double> %18, %9
60 %20 = fadd <2 x double> %18, %9
61 %21 = shufflevector <2 x double> %19, <2 x double> %20, <2 x i32> <i32 0, i32 3>
62 %22 = insertelement <2 x double> undef, double %4, i32 0
63 %23 = shufflevector <2 x double> %22, <2 x double> undef, <2 x i32> zeroinitializer
64 %24 = fdiv <2 x double> %21, %23
65 %25 = fptosi <2 x double> %24 to <2 x i32>
66 %26 = zext <2 x i32> %25 to <2 x i64>
67 %27 = shl nuw <2 x i64> %26, <i64 32, i64 32>
68 %28 = zext <2 x i32> %16 to <2 x i64>
69 %29 = or <2 x i64> %27, %28
70 %30 = extractelement <2 x i64> %29, i32 0
71 %31 = insertvalue { i64, i64 } undef, i64 %30, 0
72 %32 = extractelement <2 x i64> %29, i32 1
73 %33 = insertvalue { i64, i64 } %31, i64 %32, 1