1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -march=ve -mattr=+vpu | FileCheck %s
4 declare <256 x float> @llvm.vp.fadd.v256f32(<256 x float>, <256 x float>, <256 x i1>, i32)
6 define fastcc <256 x float> @test_vp_fadd_v256f32_vv(<256 x float> %i0, <256 x float> %i1, <256 x i1> %m, i32 %n) {
7 ; CHECK-LABEL: test_vp_fadd_v256f32_vv:
9 ; CHECK-NEXT: and %s0, %s0, (32)0
11 ; CHECK-NEXT: pvfadd.up %v0, %v0, %v1, %vm1
12 ; CHECK-NEXT: b.l.t (, %s10)
13 %r0 = call <256 x float> @llvm.vp.fadd.v256f32(<256 x float> %i0, <256 x float> %i1, <256 x i1> %m, i32 %n)
17 define fastcc <256 x float> @test_vp_fadd_v256f32_rv(float %s0, <256 x float> %i1, <256 x i1> %m, i32 %n) {
18 ; CHECK-LABEL: test_vp_fadd_v256f32_rv:
20 ; CHECK-NEXT: and %s1, %s1, (32)0
22 ; CHECK-NEXT: pvfadd.up %v0, %s0, %v0, %vm1
23 ; CHECK-NEXT: b.l.t (, %s10)
24 %xins = insertelement <256 x float> undef, float %s0, i32 0
25 %i0 = shufflevector <256 x float> %xins, <256 x float> undef, <256 x i32> zeroinitializer
26 %r0 = call <256 x float> @llvm.vp.fadd.v256f32(<256 x float> %i0, <256 x float> %i1, <256 x i1> %m, i32 %n)
30 define fastcc <256 x float> @test_vp_fadd_v256f32_vr(<256 x float> %i0, float %s1, <256 x i1> %m, i32 %n) {
31 ; CHECK-LABEL: test_vp_fadd_v256f32_vr:
33 ; CHECK-NEXT: and %s1, %s1, (32)0
35 ; CHECK-NEXT: pvfadd.up %v0, %s0, %v0, %vm1
36 ; CHECK-NEXT: b.l.t (, %s10)
37 %yins = insertelement <256 x float> undef, float %s1, i32 0
38 %i1 = shufflevector <256 x float> %yins, <256 x float> undef, <256 x i32> zeroinitializer
39 %r0 = call <256 x float> @llvm.vp.fadd.v256f32(<256 x float> %i0, <256 x float> %i1, <256 x i1> %m, i32 %n)
44 declare <256 x double> @llvm.vp.fadd.v256f64(<256 x double>, <256 x double>, <256 x i1>, i32)
46 define fastcc <256 x double> @test_vp_fadd_v256f64_vv(<256 x double> %i0, <256 x double> %i1, <256 x i1> %m, i32 %n) {
47 ; CHECK-LABEL: test_vp_fadd_v256f64_vv:
49 ; CHECK-NEXT: and %s0, %s0, (32)0
51 ; CHECK-NEXT: vfadd.d %v0, %v0, %v1, %vm1
52 ; CHECK-NEXT: b.l.t (, %s10)
53 %r0 = call <256 x double> @llvm.vp.fadd.v256f64(<256 x double> %i0, <256 x double> %i1, <256 x i1> %m, i32 %n)
54 ret <256 x double> %r0
57 define fastcc <256 x double> @test_vp_fadd_v256f64_rv(double %s0, <256 x double> %i1, <256 x i1> %m, i32 %n) {
58 ; CHECK-LABEL: test_vp_fadd_v256f64_rv:
60 ; CHECK-NEXT: and %s1, %s1, (32)0
62 ; CHECK-NEXT: vfadd.d %v0, %s0, %v0, %vm1
63 ; CHECK-NEXT: b.l.t (, %s10)
64 %xins = insertelement <256 x double> undef, double %s0, i32 0
65 %i0 = shufflevector <256 x double> %xins, <256 x double> undef, <256 x i32> zeroinitializer
66 %r0 = call <256 x double> @llvm.vp.fadd.v256f64(<256 x double> %i0, <256 x double> %i1, <256 x i1> %m, i32 %n)
67 ret <256 x double> %r0
70 define fastcc <256 x double> @test_vp_fadd_v256f64_vr(<256 x double> %i0, double %s1, <256 x i1> %m, i32 %n) {
71 ; CHECK-LABEL: test_vp_fadd_v256f64_vr:
73 ; CHECK-NEXT: and %s1, %s1, (32)0
75 ; CHECK-NEXT: vfadd.d %v0, %s0, %v0, %vm1
76 ; CHECK-NEXT: b.l.t (, %s10)
77 %yins = insertelement <256 x double> undef, double %s1, i32 0
78 %i1 = shufflevector <256 x double> %yins, <256 x double> undef, <256 x i32> zeroinitializer
79 %r0 = call <256 x double> @llvm.vp.fadd.v256f64(<256 x double> %i0, <256 x double> %i1, <256 x i1> %m, i32 %n)
80 ret <256 x double> %r0