1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple powerpc64-ibm-aix-xcoff -mcpu=pwr7 \
3 ; RUN: -verify-machineinstrs -O2 -mattr=vsx < %s | \
4 ; RUN: FileCheck %s --check-prefix=CHECK-AIX
5 ; RUN: llc -mtriple powerpc64le-unknown-linux-gnu -mcpu=pwr8 \
6 ; RUN: -verify-machineinstrs -O2 -mattr=vsx < %s | \
9 define void @test_aix_splatimm(i32 %arg, i32 %arg1, i32 %arg2) {
10 ; CHECK-AIX-LABEL: test_aix_splatimm:
11 ; CHECK-AIX: # %bb.0: # %bb
12 ; CHECK-AIX-NEXT: bclr 12, 20, 0
13 ; CHECK-AIX-NEXT: # %bb.1: # %bb3
14 ; CHECK-AIX-NEXT: slwi 3, 3, 8
15 ; CHECK-AIX-NEXT: neg 3, 3
16 ; CHECK-AIX-NEXT: lwz 6, 0(3)
17 ; CHECK-AIX-NEXT: sth 3, -16(1)
18 ; CHECK-AIX-NEXT: addi 3, 1, -16
19 ; CHECK-AIX-NEXT: lxvw4x 34, 0, 3
20 ; CHECK-AIX-NEXT: srwi 3, 4, 16
21 ; CHECK-AIX-NEXT: srwi 4, 5, 16
22 ; CHECK-AIX-NEXT: mullw 3, 4, 3
23 ; CHECK-AIX-NEXT: srwi 4, 6, 1
24 ; CHECK-AIX-NEXT: mullw 3, 3, 4
25 ; CHECK-AIX-NEXT: li 4, 0
26 ; CHECK-AIX-NEXT: neg 3, 3
27 ; CHECK-AIX-NEXT: vsplth 2, 2, 0
28 ; CHECK-AIX-NEXT: stxvw4x 34, 0, 4
29 ; CHECK-AIX-NEXT: sth 3, -32(1)
30 ; CHECK-AIX-NEXT: addi 3, 1, -32
31 ; CHECK-AIX-NEXT: lxvw4x 34, 0, 3
32 ; CHECK-AIX-NEXT: vsplth 2, 2, 0
33 ; CHECK-AIX-NEXT: stxvw4x 34, 0, 3
35 ; CHECK-LABEL: test_aix_splatimm:
36 ; CHECK: # %bb.0: # %bb
37 ; CHECK-NEXT: bclr 12, 20, 0
38 ; CHECK-NEXT: # %bb.1: # %bb3
39 ; CHECK-NEXT: slwi 3, 3, 8
40 ; CHECK-NEXT: srwi 4, 4, 16
41 ; CHECK-NEXT: neg 3, 3
42 ; CHECK-NEXT: srwi 5, 5, 16
43 ; CHECK-NEXT: mullw 4, 5, 4
44 ; CHECK-NEXT: lwz 5, 0(3)
45 ; CHECK-NEXT: mtvsrd 34, 3
47 ; CHECK-NEXT: srwi 5, 5, 1
48 ; CHECK-NEXT: mullw 4, 4, 5
49 ; CHECK-NEXT: neg 4, 4
50 ; CHECK-NEXT: mtvsrd 35, 4
51 ; CHECK-NEXT: vsplth 2, 2, 3
52 ; CHECK-NEXT: stxvd2x 34, 0, 3
53 ; CHECK-NEXT: vsplth 3, 3, 3
54 ; CHECK-NEXT: stxvd2x 35, 0, 3
56 br i1 undef, label %bb22, label %bb3
59 %i = insertelement <8 x i16> undef, i16 0, i32 0
60 %i4 = trunc i32 %arg to i16
61 %i5 = mul i16 %i4, -256
62 %i6 = insertelement <8 x i16> %i, i16 %i5, i32 1
63 %i7 = ashr i32 %arg1, 16
64 %i8 = ashr i32 %arg2, 16
65 %i9 = mul nsw i32 %i8, %i7
66 %i10 = insertelement <8 x i16> %i6, i16 0, i32 2
67 %i11 = insertelement <8 x i16> %i10, i16 0, i32 3
68 %i12 = load i32, ptr undef, align 4
69 %i13 = ashr i32 %i12, 1
70 %i14 = mul i32 %i9, %i13
71 %i15 = trunc i32 %i14 to i16
72 %i16 = sub i16 0, %i15
73 %i17 = insertelement <8 x i16> %i11, i16 %i16, i32 4
74 %i18 = insertelement <8 x i16> %i17, i16 0, i32 5
75 %i19 = bitcast <8 x i16> %i18 to <16 x i8>
76 %i20 = shufflevector <16 x i8> %i19, <16 x i8> undef, <16 x i32> <i32 2, i32 3, i32 2, i32 3, i32 2, i32 3, i32 2, i32 3, i32 2, i32 3, i32 2, i32 3, i32 2, i32 3, i32 2, i32 3>
77 store <16 x i8> %i20, ptr null, align 16
78 %i21 = shufflevector <16 x i8> %i19, <16 x i8> undef, <16 x i32> <i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9, i32 8, i32 9>
79 store <16 x i8> %i21, ptr undef, align 16