1 ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -disable-mmx -o %t -f
2 ; RUN: grep psllq %t | count 1
3 ; RUN: grep pslld %t | count 3
4 ; RUN: grep psllw %t | count 2
6 ; test vector shifts converted to proper SSE2 vector shifts when the shift
7 ; amounts are the same when using a shuffle splat.
9 define void @shift1a(<2 x i64> %val, <2 x i64>* %dst, <2 x i64> %sh) nounwind {
11 %shamt = shufflevector <2 x i64> %sh, <2 x i64> undef, <2 x i32> <i32 0, i32 0>
12 %shl = shl <2 x i64> %val, %shamt
13 store <2 x i64> %shl, <2 x i64>* %dst
17 define void @shift1b(<2 x i64> %val, <2 x i64>* %dst, <2 x i64> %sh) nounwind {
19 %shamt = shufflevector <2 x i64> %sh, <2 x i64> undef, <2 x i32> <i32 0, i32 1>
20 %shl = shl <2 x i64> %val, %shamt
21 store <2 x i64> %shl, <2 x i64>* %dst
25 define void @shift2a(<4 x i32> %val, <4 x i32>* %dst, <2 x i32> %amt) nounwind {
27 %shamt = shufflevector <2 x i32> %amt, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
28 %shl = shl <4 x i32> %val, %shamt
29 store <4 x i32> %shl, <4 x i32>* %dst
33 define void @shift2b(<4 x i32> %val, <4 x i32>* %dst, <2 x i32> %amt) nounwind {
35 %shamt = shufflevector <2 x i32> %amt, <2 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 1, i32 1>
36 %shl = shl <4 x i32> %val, %shamt
37 store <4 x i32> %shl, <4 x i32>* %dst
41 define void @shift2c(<4 x i32> %val, <4 x i32>* %dst, <2 x i32> %amt) nounwind {
43 %shamt = shufflevector <2 x i32> %amt, <2 x i32> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1>
44 %shl = shl <4 x i32> %val, %shamt
45 store <4 x i32> %shl, <4 x i32>* %dst
49 define void @shift3a(<8 x i16> %val, <8 x i16>* %dst, <8 x i16> %amt) nounwind {
51 %shamt = shufflevector <8 x i16> %amt, <8 x i16> undef, <8 x i32> <i32 6, i32 6, i32 6, i32 6, i32 6, i32 6, i32 6, i32 6>
52 %shl = shl <8 x i16> %val, %shamt
53 store <8 x i16> %shl, <8 x i16>* %dst
57 define void @shift3b(<8 x i16> %val, <8 x i16>* %dst, i16 %amt) nounwind {
59 %0 = insertelement <8 x i16> undef, i16 %amt, i32 0
60 %1 = insertelement <8 x i16> %0, i16 %amt, i32 1
61 %2 = insertelement <8 x i16> %0, i16 %amt, i32 2
62 %3 = insertelement <8 x i16> %0, i16 %amt, i32 3
63 %4 = insertelement <8 x i16> %0, i16 %amt, i32 4
64 %5 = insertelement <8 x i16> %0, i16 %amt, i32 5
65 %6 = insertelement <8 x i16> %0, i16 %amt, i32 6
66 %7 = insertelement <8 x i16> %0, i16 %amt, i32 7
67 %shl = shl <8 x i16> %val, %7
68 store <8 x i16> %shl, <8 x i16>* %dst