Run DCE after a LoopFlatten test to reduce spurious output [nfc]
[llvm-project.git] / clang / test / CodeGen / ppc-vec_ct-truncate.c
blob42542edfad71c9bdd3420d3a976307c187ed7d6e
1 // REQUIRES: powerpc-registered-target
2 // RUN: %clang_cc1 -target-feature +altivec -target-feature +vsx \
3 // RUN: -flax-vector-conversions=all -triple powerpc64-ibm-aix-xcoff -emit-llvm %s -o - \
4 // RUN: -target-cpu pwr7 | FileCheck %s
5 // RUN: %clang_cc1 -target-feature +altivec -target-feature +vsx \
6 // RUN: -flax-vector-conversions=all -triple powerpc64-unknown-linux-gnu -emit-llvm %s -o - \
7 // RUN: -target-cpu pwr8 | FileCheck %s
8 // RUN: %clang_cc1 -target-feature +altivec -target-feature +vsx \
9 // RUN: -flax-vector-conversions=all -triple powerpc64le-unknown-linux-gnu -emit-llvm %s -o - \
10 // RUN: -target-cpu pwr8 | FileCheck %s
11 // RUN: %clang_cc1 -target-feature +altivec -target-feature +vsx \
12 // RUN: -flax-vector-conversions=all -triple powerpc64-ibm-aix-xcoff -emit-llvm %s -o - \
13 // RUN: -D__XL_COMPAT_ALTIVEC__ -target-cpu pwr7 | FileCheck %s
14 // RUN: %clang_cc1 -target-feature +altivec -target-feature +vsx \
15 // RUN: -flax-vector-conversions=all -triple powerpc64le-unknown-linux-gnu -emit-llvm %s -o - \
16 // RUN: -D__XL_COMPAT_ALTIVEC__ -target-cpu pwr8 | FileCheck %s
18 #include <altivec.h>
19 vector double a1 = {-1.234e-5, 1.2345};
20 vector signed int res_vsi;
21 vector float vf1 = {0.234, 1.234, 2.345, 3.456};
22 vector signed int vsi1 = {1, 2, 3, 4};
23 vector double res_vd;
24 vector float res_vf;
25 vector signed long long res_vsll;
26 vector unsigned long long res_vull;
27 void test(void) {
28 // CHECK-LABEL: @test(
29 // CHECK-NEXT: entry:
31 res_vsi = vec_cts(a1, 31);
32 // CHECK: [[TMP0:%.*]] = load <2 x double>, ptr @a1, align 16
33 // CHECK-NEXT: fmul <2 x double> [[TMP0]], <double 0x41E0000000000000, double 0x41E0000000000000>
35 res_vsi = vec_cts(a1, 500);
36 // CHECK: [[TMP4:%.*]] = load <2 x double>, ptr @a1, align 16
37 // CHECK-NEXT: fmul <2 x double> [[TMP4]], <double 0x4130000000000000, double 0x4130000000000000>
39 res_vsi = vec_ctu(vf1, 31);
40 // CHECK: [[TMP8:%.*]] = load <4 x float>, ptr @vf1, align 16
41 // CHECK-NEXT: call <4 x i32> @llvm.ppc.altivec.vctuxs(<4 x float> [[TMP8]], i32 31)
43 res_vsi = vec_ctu(vf1, 500);
44 // CHECK: [[TMP10:%.*]] = load <4 x float>, ptr @vf1, align 16
45 // CHECK-NEXT: call <4 x i32> @llvm.ppc.altivec.vctuxs(<4 x float> [[TMP10]], i32 20)
47 res_vull = vec_ctul(vf1, 31);
48 // CHECK: [[TMP12:%.*]] = load <4 x float>, ptr @vf1, align 16
49 // CHECK-NEXT: fmul <4 x float> [[TMP12]], <float 0x41E0000000000000, float 0x41E0000000000000, float 0x41E0000000000000, float 0x41E0000000000000>
51 res_vull = vec_ctul(vf1, 500);
52 // CHECK: [[TMP21:%.*]] = load <4 x float>, ptr @vf1, align 16
53 // CHECK-NEXT: fmul <4 x float> [[TMP21]], <float 0x4130000000000000, float 0x4130000000000000, float 0x4130000000000000, float 0x4130000000000000>
55 res_vsll = vec_ctsl(vf1, 31);
56 // CHECK: [[TMP30:%.*]] = load <4 x float>, ptr @vf1, align 16
57 // CHECK-NEXT: fmul <4 x float> [[TMP30]], <float 0x41E0000000000000, float 0x41E0000000000000, float 0x41E0000000000000, float 0x41E0000000000000>
59 res_vsll = vec_ctsl(vf1, 500);
60 // CHECK: [[TMP39:%.*]] = load <4 x float>, ptr @vf1, align 16
61 // CHECK-NEXT: fmul <4 x float> [[TMP39]], <float 0x4130000000000000, float 0x4130000000000000, float 0x4130000000000000, float 0x4130000000000000>
63 res_vf = vec_ctf(vsi1, 31);
64 // CHECK: [[TMP48:%.*]] = load <4 x i32>, ptr @vsi1, align 16
65 // CHECK-NEXT: call <4 x float> @llvm.ppc.altivec.vcfsx(<4 x i32> [[TMP48]], i32 31)
67 res_vf = vec_ctf(vsi1, 500);
68 // CHECK: [[TMP50:%.*]] = load <4 x i32>, ptr @vsi1, align 16
69 // CHECK-NEXT: call <4 x float> @llvm.ppc.altivec.vcfsx(<4 x i32> [[TMP50]], i32 20)
71 res_vd = vec_ctd(vsi1, 31);
72 // CHECK: [[TMP53:%.*]] = load <4 x i32>, ptr @vsi1, align 16
73 // CHECK: [[TMP83:%.*]] = call <2 x double> @llvm.ppc.vsx.xvcvsxwdp(<4 x i32> [[TMP82:%.*]])
74 // CHECK-NEXT: fmul <2 x double> [[TMP83]], <double 0x3E00000000000000, double 0x3E00000000000000>
76 res_vd = vec_ctd(vsi1, 500);
77 // CHECK: [[TMP84:%.*]] = load <4 x i32>, ptr @vsi1, align 16
78 // CHECK: [[TMP115:%.*]] = call <2 x double> @llvm.ppc.vsx.xvcvsxwdp(<4 x i32> [[TMP114:%.*]])
79 // CHECK-NEXT: fmul <2 x double> [[TMP115]], <double 0x3EB0000000000000, double 0x3EB0000000000000>