1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s --mattr=+sve2 -o - | FileCheck %s
4 target triple = "aarch64-arm-none-eabi"
6 ; Expected to not transform as the type's minimum size is less than 128 bits.
7 define <vscale x 4 x i16> @complex_mul_v4i16(<vscale x 4 x i16> %a, <vscale x 4 x i16> %b) {
8 ; CHECK-LABEL: complex_mul_v4i16:
9 ; CHECK: // %bb.0: // %entry
10 ; CHECK-NEXT: uunpkhi z2.d, z0.s
11 ; CHECK-NEXT: uunpklo z0.d, z0.s
12 ; CHECK-NEXT: uunpkhi z3.d, z1.s
13 ; CHECK-NEXT: uunpklo z1.d, z1.s
14 ; CHECK-NEXT: ptrue p0.d
15 ; CHECK-NEXT: uzp1 z4.d, z0.d, z2.d
16 ; CHECK-NEXT: uzp2 z0.d, z0.d, z2.d
17 ; CHECK-NEXT: uzp1 z2.d, z1.d, z3.d
18 ; CHECK-NEXT: uzp2 z1.d, z1.d, z3.d
19 ; CHECK-NEXT: mul z3.d, z2.d, z0.d
20 ; CHECK-NEXT: mul z2.d, z2.d, z4.d
21 ; CHECK-NEXT: mla z3.d, p0/m, z1.d, z4.d
22 ; CHECK-NEXT: msb z0.d, p0/m, z1.d, z2.d
23 ; CHECK-NEXT: zip2 z1.d, z0.d, z3.d
24 ; CHECK-NEXT: zip1 z0.d, z0.d, z3.d
25 ; CHECK-NEXT: uzp1 z0.s, z0.s, z1.s
28 %a.deinterleaved = tail call { <vscale x 2 x i16>, <vscale x 2 x i16> } @llvm.experimental.vector.deinterleave2.nxv4i16(<vscale x 4 x i16> %a)
29 %a.real = extractvalue { <vscale x 2 x i16>, <vscale x 2 x i16> } %a.deinterleaved, 0
30 %a.imag = extractvalue { <vscale x 2 x i16>, <vscale x 2 x i16> } %a.deinterleaved, 1
31 %b.deinterleaved = tail call { <vscale x 2 x i16>, <vscale x 2 x i16> } @llvm.experimental.vector.deinterleave2.nxv4i16(<vscale x 4 x i16> %b)
32 %b.real = extractvalue { <vscale x 2 x i16>, <vscale x 2 x i16> } %b.deinterleaved, 0
33 %b.imag = extractvalue { <vscale x 2 x i16>, <vscale x 2 x i16> } %b.deinterleaved, 1
34 %0 = mul <vscale x 2 x i16> %b.imag, %a.real
35 %1 = mul <vscale x 2 x i16> %b.real, %a.imag
36 %2 = add <vscale x 2 x i16> %1, %0
37 %3 = mul <vscale x 2 x i16> %b.real, %a.real
38 %4 = mul <vscale x 2 x i16> %a.imag, %b.imag
39 %5 = sub <vscale x 2 x i16> %3, %4
40 %interleaved.vec = tail call <vscale x 4 x i16> @llvm.experimental.vector.interleave2.nxv4i16(<vscale x 2 x i16> %5, <vscale x 2 x i16> %2)
41 ret <vscale x 4 x i16> %interleaved.vec
44 ; Expected to transform
45 define <vscale x 8 x i16> @complex_mul_v8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b) {
46 ; CHECK-LABEL: complex_mul_v8i16:
47 ; CHECK: // %bb.0: // %entry
48 ; CHECK-NEXT: mov z2.h, #0 // =0x0
49 ; CHECK-NEXT: cmla z2.h, z1.h, z0.h, #0
50 ; CHECK-NEXT: cmla z2.h, z1.h, z0.h, #90
51 ; CHECK-NEXT: mov z0.d, z2.d
54 %a.deinterleaved = tail call { <vscale x 4 x i16>, <vscale x 4 x i16> } @llvm.experimental.vector.deinterleave2.nxv8i16(<vscale x 8 x i16> %a)
55 %a.real = extractvalue { <vscale x 4 x i16>, <vscale x 4 x i16> } %a.deinterleaved, 0
56 %a.imag = extractvalue { <vscale x 4 x i16>, <vscale x 4 x i16> } %a.deinterleaved, 1
57 %b.deinterleaved = tail call { <vscale x 4 x i16>, <vscale x 4 x i16> } @llvm.experimental.vector.deinterleave2.nxv8i16(<vscale x 8 x i16> %b)
58 %b.real = extractvalue { <vscale x 4 x i16>, <vscale x 4 x i16> } %b.deinterleaved, 0
59 %b.imag = extractvalue { <vscale x 4 x i16>, <vscale x 4 x i16> } %b.deinterleaved, 1
60 %0 = mul <vscale x 4 x i16> %b.imag, %a.real
61 %1 = mul <vscale x 4 x i16> %b.real, %a.imag
62 %2 = add <vscale x 4 x i16> %1, %0
63 %3 = mul <vscale x 4 x i16> %b.real, %a.real
64 %4 = mul <vscale x 4 x i16> %a.imag, %b.imag
65 %5 = sub <vscale x 4 x i16> %3, %4
66 %interleaved.vec = tail call <vscale x 8 x i16> @llvm.experimental.vector.interleave2.nxv8i16(<vscale x 4 x i16> %5, <vscale x 4 x i16> %2)
67 ret <vscale x 8 x i16> %interleaved.vec
69 ; Expected to transform
70 define <vscale x 16 x i16> @complex_mul_v16i16(<vscale x 16 x i16> %a, <vscale x 16 x i16> %b) {
71 ; CHECK-LABEL: complex_mul_v16i16:
72 ; CHECK: // %bb.0: // %entry
73 ; CHECK-NEXT: mov z4.h, #0 // =0x0
74 ; CHECK-NEXT: mov z5.d, z4.d
75 ; CHECK-NEXT: cmla z4.h, z3.h, z1.h, #0
76 ; CHECK-NEXT: cmla z5.h, z2.h, z0.h, #0
77 ; CHECK-NEXT: cmla z4.h, z3.h, z1.h, #90
78 ; CHECK-NEXT: cmla z5.h, z2.h, z0.h, #90
79 ; CHECK-NEXT: mov z1.d, z4.d
80 ; CHECK-NEXT: mov z0.d, z5.d
83 %a.deinterleaved = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.experimental.vector.deinterleave2.nxv16i16(<vscale x 16 x i16> %a)
84 %a.real = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %a.deinterleaved, 0
85 %a.imag = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %a.deinterleaved, 1
86 %b.deinterleaved = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.experimental.vector.deinterleave2.nxv16i16(<vscale x 16 x i16> %b)
87 %b.real = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %b.deinterleaved, 0
88 %b.imag = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %b.deinterleaved, 1
89 %0 = mul <vscale x 8 x i16> %b.imag, %a.real
90 %1 = mul <vscale x 8 x i16> %b.real, %a.imag
91 %2 = add <vscale x 8 x i16> %1, %0
92 %3 = mul <vscale x 8 x i16> %b.real, %a.real
93 %4 = mul <vscale x 8 x i16> %a.imag, %b.imag
94 %5 = sub <vscale x 8 x i16> %3, %4
95 %interleaved.vec = tail call <vscale x 16 x i16> @llvm.experimental.vector.interleave2.nxv16i16(<vscale x 8 x i16> %5, <vscale x 8 x i16> %2)
96 ret <vscale x 16 x i16> %interleaved.vec
99 ; Expected to transform
100 define <vscale x 32 x i16> @complex_mul_v32i16(<vscale x 32 x i16> %a, <vscale x 32 x i16> %b) {
101 ; CHECK-LABEL: complex_mul_v32i16:
102 ; CHECK: // %bb.0: // %entry
103 ; CHECK-NEXT: mov z24.h, #0 // =0x0
104 ; CHECK-NEXT: mov z25.d, z24.d
105 ; CHECK-NEXT: mov z26.d, z24.d
106 ; CHECK-NEXT: mov z27.d, z24.d
107 ; CHECK-NEXT: cmla z24.h, z7.h, z3.h, #0
108 ; CHECK-NEXT: cmla z25.h, z4.h, z0.h, #0
109 ; CHECK-NEXT: cmla z26.h, z5.h, z1.h, #0
110 ; CHECK-NEXT: cmla z27.h, z6.h, z2.h, #0
111 ; CHECK-NEXT: cmla z24.h, z7.h, z3.h, #90
112 ; CHECK-NEXT: cmla z25.h, z4.h, z0.h, #90
113 ; CHECK-NEXT: cmla z26.h, z5.h, z1.h, #90
114 ; CHECK-NEXT: cmla z27.h, z6.h, z2.h, #90
115 ; CHECK-NEXT: mov z3.d, z24.d
116 ; CHECK-NEXT: mov z0.d, z25.d
117 ; CHECK-NEXT: mov z1.d, z26.d
118 ; CHECK-NEXT: mov z2.d, z27.d
121 %a.deinterleaved = tail call { <vscale x 16 x i16>, <vscale x 16 x i16> } @llvm.experimental.vector.deinterleave2.nxv32i16(<vscale x 32 x i16> %a)
122 %a.real = extractvalue { <vscale x 16 x i16>, <vscale x 16 x i16> } %a.deinterleaved, 0
123 %a.imag = extractvalue { <vscale x 16 x i16>, <vscale x 16 x i16> } %a.deinterleaved, 1
124 %b.deinterleaved = tail call { <vscale x 16 x i16>, <vscale x 16 x i16> } @llvm.experimental.vector.deinterleave2.nxv32i16(<vscale x 32 x i16> %b)
125 %b.real = extractvalue { <vscale x 16 x i16>, <vscale x 16 x i16> } %b.deinterleaved, 0
126 %b.imag = extractvalue { <vscale x 16 x i16>, <vscale x 16 x i16> } %b.deinterleaved, 1
127 %0 = mul <vscale x 16 x i16> %b.imag, %a.real
128 %1 = mul <vscale x 16 x i16> %b.real, %a.imag
129 %2 = add <vscale x 16 x i16> %1, %0
130 %3 = mul <vscale x 16 x i16> %b.real, %a.real
131 %4 = mul <vscale x 16 x i16> %a.imag, %b.imag
132 %5 = sub <vscale x 16 x i16> %3, %4
133 %interleaved.vec = tail call <vscale x 32 x i16> @llvm.experimental.vector.interleave2.nxv32i16(<vscale x 16 x i16> %5, <vscale x 16 x i16> %2)
134 ret <vscale x 32 x i16> %interleaved.vec
137 declare { <vscale x 2 x i16>, <vscale x 2 x i16> } @llvm.experimental.vector.deinterleave2.nxv4i16(<vscale x 4 x i16>)
138 declare <vscale x 4 x i16> @llvm.experimental.vector.interleave2.nxv4i16(<vscale x 2 x i16>, <vscale x 2 x i16>)
140 declare { <vscale x 4 x i16>, <vscale x 4 x i16> } @llvm.experimental.vector.deinterleave2.nxv8i16(<vscale x 8 x i16>)
141 declare <vscale x 8 x i16> @llvm.experimental.vector.interleave2.nxv8i16(<vscale x 4 x i16>, <vscale x 4 x i16>)
143 declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.experimental.vector.deinterleave2.nxv16i16(<vscale x 16 x i16>)
144 declare <vscale x 16 x i16> @llvm.experimental.vector.interleave2.nxv16i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
146 declare { <vscale x 16 x i16>, <vscale x 16 x i16> } @llvm.experimental.vector.deinterleave2.nxv32i16(<vscale x 32 x i16>)
147 declare <vscale x 32 x i16> @llvm.experimental.vector.interleave2.nxv32i16(<vscale x 16 x i16>, <vscale x 16 x i16>)