1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -verify-machineinstrs < %s | FileCheck %s
4 target triple = "aarch64-linux-gnu"
6 define <vscale x 16 x i8> @test_sclamp_i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c) #0 {
7 ; CHECK-LABEL: test_sclamp_i8:
9 ; CHECK-NEXT: sclamp z2.b, z0.b, z1.b
10 ; CHECK-NEXT: mov z0.d, z2.d
12 %res = call <vscale x 16 x i8> @llvm.aarch64.sve.sclamp.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c)
13 ret <vscale x 16 x i8> %res
16 define <vscale x 8 x i16> @test_sclamp_i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c) #0 {
17 ; CHECK-LABEL: test_sclamp_i16:
19 ; CHECK-NEXT: sclamp z2.h, z0.h, z1.h
20 ; CHECK-NEXT: mov z0.d, z2.d
22 %res = call <vscale x 8 x i16> @llvm.aarch64.sve.sclamp.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c)
23 ret <vscale x 8 x i16> %res
26 define <vscale x 4 x i32> @test_sclamp_i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c) #0 {
27 ; CHECK-LABEL: test_sclamp_i32:
29 ; CHECK-NEXT: sclamp z2.s, z0.s, z1.s
30 ; CHECK-NEXT: mov z0.d, z2.d
32 %res = call <vscale x 4 x i32> @llvm.aarch64.sve.sclamp.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c)
33 ret <vscale x 4 x i32> %res
36 define <vscale x 2 x i64> @test_sclamp_i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c) #0 {
37 ; CHECK-LABEL: test_sclamp_i64:
39 ; CHECK-NEXT: sclamp z2.d, z0.d, z1.d
40 ; CHECK-NEXT: mov z0.d, z2.d
42 %res = call <vscale x 2 x i64> @llvm.aarch64.sve.sclamp.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c)
43 ret <vscale x 2 x i64> %res
46 define { <vscale x 16 x i8>, <vscale x 16 x i8> } @test_sclamp_single_x2_i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c, <vscale x 16 x i8> %d) #1 {
47 ; CHECK-LABEL: test_sclamp_single_x2_i8:
49 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
50 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
51 ; CHECK-NEXT: sclamp { z0.b, z1.b }, z2.b, z3.b
53 %res = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.sclamp.single.x2.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c, <vscale x 16 x i8> %d)
54 ret { <vscale x 16 x i8>, <vscale x 16 x i8> } %res
57 define { <vscale x 8 x i16>, <vscale x 8 x i16> } @test_sclamp_single_x2_i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c, <vscale x 8 x i16> %d) #1 {
58 ; CHECK-LABEL: test_sclamp_single_x2_i16:
60 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
61 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
62 ; CHECK-NEXT: sclamp { z0.h, z1.h }, z2.h, z3.h
64 %res = call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.sclamp.single.x2.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c, <vscale x 8 x i16> %d)
65 ret { <vscale x 8 x i16>, <vscale x 8 x i16> } %res
68 define { <vscale x 4 x i32>, <vscale x 4 x i32> } @test_sclamp_single_x2_i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c, <vscale x 4 x i32> %d) #1 {
69 ; CHECK-LABEL: test_sclamp_single_x2_i32:
71 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
72 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
73 ; CHECK-NEXT: sclamp { z0.s, z1.s }, z2.s, z3.s
75 %res = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.sclamp.single.x2.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c, <vscale x 4 x i32> %d)
76 ret { <vscale x 4 x i32>, <vscale x 4 x i32> } %res
79 define { <vscale x 2 x i64>, <vscale x 2 x i64> } @test_sclamp_single_x2_i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c, <vscale x 2 x i64> %d) #1 {
80 ; CHECK-LABEL: test_sclamp_single_x2_i64:
82 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
83 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
84 ; CHECK-NEXT: sclamp { z0.d, z1.d }, z2.d, z3.d
86 %res = call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.sclamp.single.x2.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c, <vscale x 2 x i64> %d)
87 ret { <vscale x 2 x i64>, <vscale x 2 x i64> } %res
90 define { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @test_sclamp_single_x4_i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c, <vscale x 16 x i8> %d, <vscale x 16 x i8> %e, <vscale x 16 x i8> %f) #1 {
91 ; CHECK-LABEL: test_sclamp_single_x4_i8:
93 ; CHECK-NEXT: // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
94 ; CHECK-NEXT: // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
95 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
96 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
97 ; CHECK-NEXT: sclamp { z0.b - z3.b }, z4.b, z5.b
99 %res = call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.sclamp.single.x4.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c, <vscale x 16 x i8> %d, <vscale x 16 x i8> %e, <vscale x 16 x i8> %f)
100 ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %res
103 define { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @test_sclamp_single_x4_i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c, <vscale x 8 x i16> %d, <vscale x 8 x i16> %e, <vscale x 8 x i16> %f) #1 {
104 ; CHECK-LABEL: test_sclamp_single_x4_i16:
106 ; CHECK-NEXT: // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
107 ; CHECK-NEXT: // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
108 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
109 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
110 ; CHECK-NEXT: sclamp { z0.h - z3.h }, z4.h, z5.h
112 %res = call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.sclamp.single.x4.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c, <vscale x 8 x i16> %d, <vscale x 8 x i16> %e, <vscale x 8 x i16> %f)
113 ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %res
116 define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @test_sclamp_single_x4_i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c, <vscale x 4 x i32> %d, <vscale x 4 x i32> %e, <vscale x 4 x i32> %f) #1 {
117 ; CHECK-LABEL: test_sclamp_single_x4_i32:
119 ; CHECK-NEXT: // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
120 ; CHECK-NEXT: // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
121 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
122 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
123 ; CHECK-NEXT: sclamp { z0.s - z3.s }, z4.s, z5.s
125 %res = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.sclamp.single.x4.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c, <vscale x 4 x i32> %d, <vscale x 4 x i32> %e, <vscale x 4 x i32> %f)
126 ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %res
129 define { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @test_sclamp_single_x4_i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c, <vscale x 2 x i64> %d, <vscale x 2 x i64> %e, <vscale x 2 x i64> %f) #1 {
130 ; CHECK-LABEL: test_sclamp_single_x4_i64:
132 ; CHECK-NEXT: // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
133 ; CHECK-NEXT: // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
134 ; CHECK-NEXT: // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
135 ; CHECK-NEXT: // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
136 ; CHECK-NEXT: sclamp { z0.d - z3.d }, z4.d, z5.d
138 %res = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.sclamp.single.x4.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c, <vscale x 2 x i64> %d, <vscale x 2 x i64> %e, <vscale x 2 x i64> %f)
139 ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %res
143 attributes #0 = { "target-features"="+sve2p1" }
144 attributes #1 = { "target-features"="+sme2" }
146 declare <vscale x 16 x i8> @llvm.aarch64.sve.sclamp.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>)
147 declare <vscale x 8 x i16> @llvm.aarch64.sve.sclamp.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>)
148 declare <vscale x 4 x i32> @llvm.aarch64.sve.sclamp.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
149 declare <vscale x 2 x i64> @llvm.aarch64.sve.sclamp.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
151 declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.sclamp.single.x2.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>)
152 declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.sclamp.single.x2.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>)
153 declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.sclamp.single.x2.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
154 declare { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.sclamp.single.x2.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
156 declare { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.sclamp.single.x4.nxv16i8(<vscale x 16 x i8> %a, <vscale x 16 x i8> %b, <vscale x 16 x i8> %c, <vscale x 16 x i8> %d, <vscale x 16 x i8> %e, <vscale x 16 x i8> %f)
157 declare { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.sclamp.single.x4.nxv8i16(<vscale x 8 x i16> %a, <vscale x 8 x i16> %b, <vscale x 8 x i16> %c, <vscale x 8 x i16> %d, <vscale x 8 x i16> %e, <vscale x 8 x i16> %f)
158 declare { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.sclamp.single.x4.nxv4i32(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b, <vscale x 4 x i32> %c, <vscale x 4 x i32> %d, <vscale x 4 x i32> %e, <vscale x 4 x i32> %f)
159 declare { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.sclamp.single.x4.nxv2i64(<vscale x 2 x i64> %a, <vscale x 2 x i64> %b, <vscale x 2 x i64> %c, <vscale x 2 x i64> %d, <vscale x 2 x i64> %e, <vscale x 2 x i64> %f)