main/gcc/testsuite/gcc.dg/vect/slp-10.c

   1 /* { dg-require-effective-target vect_int } */
   2
   3 #include <stdarg.h>
   4 #include "tree-vect.h"
   5
   6 #define N 8
   7
   8 int
   9 main1 ()
  10 {
  11   int i;
  12   unsigned int out[N*8], a0, a1, a2, a3, a4, a5, a6, a7, b1, b0, b2, b3, b4, b5, b6, b7;
  13   unsigned int in[N*8] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63};
  14   float out2[N*8];
  15
  16   for (i = 0; i < N; i++)
  17     {
  18
  19       a0 = in[i*8] + 5;
  20       a1 = in[i*8 + 1] + 6;
  21       a2 = in[i*8 + 2] + 7;
  22       a3 = in[i*8 + 3] + 8;
  23       a4 = in[i*8 + 4] + 9;
  24       a5 = in[i*8 + 5] + 10;
  25       a6 = in[i*8 + 6] + 11;
  26       a7 = in[i*8 + 7] + 12;
  27
  28       b0 = a0 * 3;
  29       b1 = a1 * 2;
  30       b2 = a2 * 12;
  31       b3 = a3 * 5;
  32       b4 = a4 * 8;
  33       b5 = a5 * 4;
  34       b6 = a6 * 3;
  35       b7 = a7 * 2;
  36
  37       out[i*8] = b0 - 2;
  38       out[i*8 + 1] = b1 - 3;
  39       out[i*8 + 2] = b2 - 2;
  40       out[i*8 + 3] = b3 - 1;
  41       out[i*8 + 4] = b4 - 8;
  42       out[i*8 + 5] = b5 - 7;
  43       out[i*8 + 6] = b6 - 3;
  44       out[i*8 + 7] = b7 - 7;
  45     }
  46
  47   /* check results:  */
  48   for (i = 0; i < N; i++)
  49     {
  50       if (out[i*8] !=  (in[i*8] + 5) * 3 - 2
  51          || out[i*8 + 1] != (in[i*8 + 1] + 6) * 2 - 3
  52          || out[i*8 + 2] != (in[i*8 + 2] + 7) * 12 - 2
  53          || out[i*8 + 3] != (in[i*8 + 3] + 8) * 5 - 1
  54          || out[i*8 + 4] != (in[i*8 + 4] + 9) * 8 - 8
  55          || out[i*8 + 5] != (in[i*8 + 5] + 10) * 4 - 7
  56          || out[i*8 + 6] != (in[i*8 + 6] + 11) * 3 - 3
  57          || out[i*8 + 7] != (in[i*8 + 7] + 12) * 2 - 7)
  58         abort ();
  59     }
  60
  61   for (i = 0; i < N*2; i++)
  62     {
  63       out[i*4] = (in[i*4] + 2) * 3;
  64       out[i*4 + 1] = (in[i*4 + 1] + 2) * 7;
  65       out[i*4 + 2] = (in[i*4 + 2] + 7) * 3;
  66       out[i*4 + 3] = (in[i*4 + 3] + 7) * 7;
  67     }
  68
  69   /* check results:  */
  70   for (i = 0; i < N*2; i++)
  71     {
  72       if (out[i*4] !=  (in[i*4] + 2) * 3
  73          || out[i*4 + 1] != (in[i*4 + 1] + 2) * 7
  74          || out[i*4 + 2] != (in[i*4 + 2] + 7) * 3
  75          || out[i*4 + 3] != (in[i*4 + 3] + 7) * 7)
  76         abort ();
  77     }
  78
  79   for (i = 0; i < N*4; i++)
  80     {
  81       out2[i*2] = (float) (in[i*2] * 2 + 5) ;
  82       out2[i*2 + 1] = (float) (in[i*2 + 1] * 3 + 7);
  83     }
  84
  85   /* check results:  */
  86   for (i = 0; i < N*4; i++)
  87     {
  88       if (out2[i*2] !=  (float) (in[i*2] * 2 + 5)
  89          || out2[i*2 + 1] != (float) (in[i*2 + 1] * 3 + 7))
  90         abort ();
  91     }
  92
  93
  94   return 0;
  95 }
  96
  97 int main (void)
  98 {
  99   check_vect ();
 100
 101   main1 ();
 102
 103   return 0;
 104 }
 105
 106 /* { dg-final { scan-tree-dump-times "vectorized 3 loops" 1 "vect"  {target {vect_uintfloat_cvt && vect_int_mult} } } } */
 107 /* { dg-final { scan-tree-dump-times "vectorized 2 loops" 1 "vect"  {target {{! { vect_uintfloat_cvt}} && vect_int_mult} } } } */
 108 /* { dg-final { scan-tree-dump-times "vectorized 0 loops" 1 "vect"  {target {{! { vect_uintfloat_cvt}} && { ! {vect_int_mult}}} } } } */
 109 /* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 3 "vect" {target {vect_uintfloat_cvt && vect_int_mult} } } } */
 110 /* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 2 "vect"  {target {{! { vect_uintfloat_cvt}} && vect_int_mult} } } } */
 111 /* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 0 "vect"  {target {{! { vect_uintfloat_cvt}} && { ! {vect_int_mult}}} } } } */
 112 /* { dg-final { cleanup-tree-dump "vect" } } */
 113