clang/test/CodeGen/SystemZ/builtins-systemz-vector2.c

   1 // REQUIRES: systemz-registered-target
   2 // RUN: %clang_cc1 -target-cpu z14 -triple s390x-ibm-linux -flax-vector-conversions=none \
   3 // RUN: -Wall -Wno-unused -Werror -emit-llvm %s -o - | FileCheck %s
   4
   5 typedef __attribute__((vector_size(16))) signed char vec_schar;
   6 typedef __attribute__((vector_size(16))) signed short vec_sshort;
   7 typedef __attribute__((vector_size(16))) signed int vec_sint;
   8 typedef __attribute__((vector_size(16))) signed long long vec_slong;
   9 typedef __attribute__((vector_size(16))) unsigned char vec_uchar;
  10 typedef __attribute__((vector_size(16))) unsigned short vec_ushort;
  11 typedef __attribute__((vector_size(16))) unsigned int vec_uint;
  12 typedef __attribute__((vector_size(16))) unsigned long long vec_ulong;
  13 typedef __attribute__((vector_size(16))) double vec_double;
  14 typedef __attribute__((vector_size(16))) float vec_float;
  15
  16 volatile vec_schar vsc;
  17 volatile vec_sshort vss;
  18 volatile vec_sint vsi;
  19 volatile vec_slong vsl;
  20 volatile vec_uchar vuc;
  21 volatile vec_ushort vus;
  22 volatile vec_uint vui;
  23 volatile vec_ulong vul;
  24 volatile vec_double vd;
  25 volatile vec_float vf;
  26 volatile unsigned __int128 ui128;
  27
  28 volatile unsigned int len;
  29 const void * volatile cptr;
  30 void * volatile ptr;
  31 int cc;
  32
  33 void test_core(void) {
  34   vul = __builtin_s390_vbperm(vuc, vuc);
  35   // CHECK: call <2 x i64> @llvm.s390.vbperm(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
  36
  37   vsc = __builtin_s390_vlrlr(len, cptr);
  38   // CHECK: call <16 x i8> @llvm.s390.vlrl(i32 %{{.*}}, ptr %{{.*}})
  39
  40   __builtin_s390_vstrlr(vsc, len, ptr);
  41   // CHECK: call void @llvm.s390.vstrl(<16 x i8> %{{.*}}, i32 %{{.*}}, ptr %{{.*}})
  42 }
  43
  44 void test_integer(void) {
  45   ui128 = __builtin_s390_vmslg(vul, vul, ui128, 0);
  46   // CHECK: call i128 @llvm.s390.vmslg(<2 x i64> %{{.*}}, <2 x i64> %{{.*}}, i128 %{{.*}}, i32 0)
  47   ui128 = __builtin_s390_vmslg(vul, vul, ui128, 15);
  48   // CHECK: call i128 @llvm.s390.vmslg(<2 x i64> %{{.*}}, <2 x i64> %{{.*}}, i128 %{{.*}}, i32 15)
  49 }
  50
  51 void test_float(void) {
  52   vd = __builtin_s390_vfmaxdb(vd, vd, 4);
  53   // CHECK: call <2 x double> @llvm.maxnum.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}})
  54   vd = __builtin_s390_vfmaxdb(vd, vd, 0);
  55   // CHECK: call <2 x double> @llvm.s390.vfmaxdb(<2 x double> %{{.*}}, <2 x double> %{{.*}}, i32 0)
  56   vd = __builtin_s390_vfmaxdb(vd, vd, 15);
  57   // CHECK: call <2 x double> @llvm.s390.vfmaxdb(<2 x double> %{{.*}}, <2 x double> %{{.*}}, i32 15)
  58
  59   vd = __builtin_s390_vfmindb(vd, vd, 4);
  60   // CHECK: call <2 x double> @llvm.minnum.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}})
  61   vd = __builtin_s390_vfmindb(vd, vd, 0);
  62   // CHECK: call <2 x double> @llvm.s390.vfmindb(<2 x double> %{{.*}}, <2 x double> %{{.*}}, i32 0)
  63   vd = __builtin_s390_vfmindb(vd, vd, 15);
  64   // CHECK: call <2 x double> @llvm.s390.vfmindb(<2 x double> %{{.*}}, <2 x double> %{{.*}}, i32 15)
  65
  66   vd = __builtin_s390_vfnmadb(vd, vd, vd);
  67   // CHECK: [[RES:%[^ ]+]] = call <2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> %{{.*}})
  68   // CHECK: fneg <2 x double> [[RES]]
  69   vd = __builtin_s390_vfnmsdb(vd, vd, vd);
  70   // CHECK: [[NEG:%[^ ]+]] = fneg <2 x double> %{{.*}}
  71   // CHECK: [[RES:%[^ ]+]] = call <2 x double> @llvm.fma.v2f64(<2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x double> [[NEG]])
  72   // CHECK: fneg <2 x double> [[RES]]
  73
  74   vsi = __builtin_s390_vfcesbs(vf, vf, &cc);
  75   // CHECK: call { <4 x i32>, i32 } @llvm.s390.vfcesbs(<4 x float> %{{.*}}, <4 x float> %{{.*}})
  76   vsi = __builtin_s390_vfchsbs(vf, vf, &cc);
  77   // CHECK: call { <4 x i32>, i32 } @llvm.s390.vfchsbs(<4 x float> %{{.*}}, <4 x float> %{{.*}})
  78   vsi = __builtin_s390_vfchesbs(vf, vf, &cc);
  79   // CHECK: call { <4 x i32>, i32 } @llvm.s390.vfchesbs(<4 x float> %{{.*}}, <4 x float> %{{.*}})
  80
  81   vsi = __builtin_s390_vftcisb(vf, 0, &cc);
  82   // CHECK: call { <4 x i32>, i32 } @llvm.s390.vftcisb(<4 x float> %{{.*}}, i32 0)
  83   vsi = __builtin_s390_vftcisb(vf, 4095, &cc);
  84   // CHECK: call { <4 x i32>, i32 } @llvm.s390.vftcisb(<4 x float> %{{.*}}, i32 4095)
  85
  86   vf = __builtin_s390_vfmaxsb(vf, vf, 4);
  87   // CHECK: call <4 x float> @llvm.maxnum.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}})
  88   vf = __builtin_s390_vfmaxsb(vf, vf, 0);
  89   // CHECK: call <4 x float> @llvm.s390.vfmaxsb(<4 x float> %{{.*}}, <4 x float> %{{.*}}, i32 0)
  90   vf = __builtin_s390_vfmaxsb(vf, vf, 15);
  91   // CHECK: call <4 x float> @llvm.s390.vfmaxsb(<4 x float> %{{.*}}, <4 x float> %{{.*}}, i32 15)
  92
  93   vf = __builtin_s390_vfminsb(vf, vf, 4);
  94   // CHECK: call <4 x float> @llvm.minnum.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}})
  95   vf = __builtin_s390_vfminsb(vf, vf, 0);
  96   // CHECK: call <4 x float> @llvm.s390.vfminsb(<4 x float> %{{.*}}, <4 x float> %{{.*}}, i32 0)
  97   vf = __builtin_s390_vfminsb(vf, vf, 15);
  98   // CHECK: call <4 x float> @llvm.s390.vfminsb(<4 x float> %{{.*}}, <4 x float> %{{.*}}, i32 15)
  99
 100   vf = __builtin_s390_vfsqsb(vf);
 101   // CHECK: call <4 x float> @llvm.sqrt.v4f32(<4 x float> %{{.*}})
 102
 103   vf = __builtin_s390_vfmasb(vf, vf, vf);
 104   // CHECK: call <4 x float> @llvm.fma.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x float> %{{.*}})
 105   vf = __builtin_s390_vfmssb(vf, vf, vf);
 106   // CHECK: [[NEG:%[^ ]+]] = fneg <4 x float> %{{.*}}
 107   // CHECK: call <4 x float> @llvm.fma.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x float> [[NEG]])
 108   vf = __builtin_s390_vfnmasb(vf, vf, vf);
 109   // CHECK: [[RES:%[^ ]+]] = call <4 x float> @llvm.fma.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x float> %{{.*}})
 110   // CHECK: fneg <4 x float> [[RES]]
 111   vf = __builtin_s390_vfnmssb(vf, vf, vf);
 112   // CHECK: [[NEG:%[^ ]+]] = fneg <4 x float> %{{.*}}
 113   // CHECK: [[RES:%[^ ]+]] = call <4 x float> @llvm.fma.v4f32(<4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x float> [[NEG]])
 114   // CHECK: fneg <4 x float> [[RES]]
 115
 116   vf = __builtin_s390_vflpsb(vf);
 117   // CHECK: call <4 x float> @llvm.fabs.v4f32(<4 x float> %{{.*}})
 118   vf = __builtin_s390_vflnsb(vf);
 119   // CHECK: [[ABS:%[^ ]+]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> %{{.*}})
 120   // CHECK: fneg <4 x float> [[ABS]]
 121
 122   vf = __builtin_s390_vfisb(vf, 0, 0);
 123   // CHECK: call <4 x float> @llvm.rint.v4f32(<4 x float> %{{.*}})
 124   vf = __builtin_s390_vfisb(vf, 4, 0);
 125   // CHECK: call <4 x float> @llvm.nearbyint.v4f32(<4 x float> %{{.*}})
 126   vf = __builtin_s390_vfisb(vf, 4, 1);
 127   // CHECK: call <4 x float> @llvm.round.v4f32(<4 x float> %{{.*}})
 128   vf = __builtin_s390_vfisb(vf, 4, 5);
 129   // CHECK: call <4 x float> @llvm.trunc.v4f32(<4 x float> %{{.*}})
 130   vf = __builtin_s390_vfisb(vf, 4, 6);
 131   // CHECK: call <4 x float> @llvm.ceil.v4f32(<4 x float> %{{.*}})
 132   vf = __builtin_s390_vfisb(vf, 4, 7);
 133   // CHECK: call <4 x float> @llvm.floor.v4f32(<4 x float> %{{.*}})
 134   vf = __builtin_s390_vfisb(vf, 4, 4);
 135   // CHECK: call <4 x float> @llvm.s390.vfisb(<4 x float> %{{.*}}, i32 4, i32 4)
 136 }
 137