1 ; RUN: llc < %s -march=x86 -mattr=+sse4.2 | FileCheck %s
3 target triple = "x86_64-unknown-unknown"
5 ; widening shuffle v3float and then a add
6 define void @shuf(<3 x float>* %dst.addr, <3 x float> %src1,<3 x float> %src2) nounwind {
8 ; CHECK: # BB#0: # %entry
9 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
10 ; CHECK-NEXT: addps %xmm1, %xmm0
11 ; CHECK-NEXT: extractps $2, %xmm0, 8(%eax)
12 ; CHECK-NEXT: extractps $1, %xmm0, 4(%eax)
13 ; CHECK-NEXT: movss %xmm0, (%eax)
16 %x = shufflevector <3 x float> %src1, <3 x float> %src2, <3 x i32> < i32 0, i32 1, i32 2>
17 %val = fadd <3 x float> %x, %src2
18 store <3 x float> %val, <3 x float>* %dst.addr
23 ; widening shuffle v3float with a different mask and then a add
24 define void @shuf2(<3 x float>* %dst.addr, <3 x float> %src1,<3 x float> %src2) nounwind {
26 ; CHECK: # BB#0: # %entry
27 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
28 ; CHECK-NEXT: blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
29 ; CHECK-NEXT: addps %xmm1, %xmm0
30 ; CHECK-NEXT: extractps $2, %xmm0, 8(%eax)
31 ; CHECK-NEXT: extractps $1, %xmm0, 4(%eax)
32 ; CHECK-NEXT: movss %xmm0, (%eax)
35 %x = shufflevector <3 x float> %src1, <3 x float> %src2, <3 x i32> < i32 0, i32 4, i32 2>
36 %val = fadd <3 x float> %x, %src2
37 store <3 x float> %val, <3 x float>* %dst.addr
41 ; Example of when widening a v3float operation causes the DAG to replace a node
42 ; with the operation that we are currently widening, i.e. when replacing
43 ; opA with opB, the DAG will produce new operations with opA.
44 define void @shuf3(<4 x float> %tmp10, <4 x float> %vecinit15, <4 x float>* %dst) nounwind {
46 ; CHECK: # BB#0: # %entry
47 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
48 ; CHECK-NEXT: pshufd {{.*#+}} xmm0 = xmm1[0,0,0,0]
49 ; CHECK-NEXT: movdqa %xmm0, (%eax)
52 %shuffle.i.i.i12 = shufflevector <4 x float> %tmp10, <4 x float> %vecinit15, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
53 %tmp25.i.i = shufflevector <4 x float> %shuffle.i.i.i12, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
54 %tmp1.i.i = shufflevector <3 x float> %tmp25.i.i, <3 x float> zeroinitializer, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
55 %tmp3.i13 = shufflevector <4 x float> %tmp1.i.i, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2> ; <<3 x float>>
56 %tmp6.i14 = shufflevector <3 x float> %tmp3.i13, <3 x float> zeroinitializer, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
57 %tmp97.i = shufflevector <4 x float> %tmp6.i14, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
58 %tmp2.i18 = shufflevector <3 x float> %tmp97.i, <3 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 2>
59 %t5 = bitcast <4 x float> %tmp2.i18 to <4 x i32>
60 %shr.i.i19 = lshr <4 x i32> %t5, <i32 19, i32 19, i32 19, i32 19>
61 %and.i.i20 = and <4 x i32> %shr.i.i19, <i32 4080, i32 4080, i32 4080, i32 4080>
62 %shuffle.i.i.i21 = shufflevector <4 x float> %tmp2.i18, <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 2, i32 3>
63 store <4 x float> %shuffle.i.i.i21, <4 x float>* %dst
67 ; PR10421: make sure we correctly handle extreme widening with CONCAT_VECTORS
68 define <8 x i8> @shuf4(<4 x i8> %a, <4 x i8> %b) nounwind readnone {
71 ; CHECK-NEXT: pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0],zero,xmm1[4],zero,xmm1[8],zero,xmm1[12],zero
72 ; CHECK-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0],zero,xmm0[4],zero,xmm0[8],zero,xmm0[12],zero,zero,zero,zero,zero,zero,zero,zero,zero
73 ; CHECK-NEXT: por %xmm1, %xmm0
75 %vshuf = shufflevector <4 x i8> %a, <4 x i8> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
79 ; PR11389: another CONCAT_VECTORS case
80 define void @shuf5(<8 x i8>* %p) nounwind {
83 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
84 ; CHECK-NEXT: movdqa {{.*#+}} xmm0 = <4,33,u,u,u,u,u,u>
85 ; CHECK-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,2,0,0,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero
86 ; CHECK-NEXT: movlpd %xmm0, (%eax)
88 %v = shufflevector <2 x i8> <i8 4, i8 33>, <2 x i8> undef, <8 x i32> <i32 1, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
89 store <8 x i8> %v, <8 x i8>* %p, align 8