2017-04-10 11:32:00 +00:00
|
|
|
/* This code uses nvptx inline assembly guarded with acc_on_device, which is
|
|
|
|
not optimized away at -O0, and then confuses the target assembler.
|
|
|
|
{ dg-skip-if "" { *-*-* } { "-O0" } { "" } } */
|
|
|
|
|
|
|
|
#include <stdio.h>
|
|
|
|
|
|
|
|
#define N (32*32*32+17)
|
|
|
|
int main ()
|
|
|
|
{
|
|
|
|
int ix;
|
|
|
|
int ondev = 0;
|
|
|
|
int t = 0, h = 0;
|
|
|
|
|
2018-12-28 15:30:48 +00:00
|
|
|
#pragma acc parallel num_gangs(32) copy(ondev)
|
2017-04-10 11:32:00 +00:00
|
|
|
{
|
|
|
|
#pragma acc loop gang reduction (+:t)
|
|
|
|
for (unsigned ix = 0; ix < N; ix++)
|
|
|
|
{
|
|
|
|
int val = ix;
|
|
|
|
|
|
|
|
if (__builtin_acc_on_device (5))
|
|
|
|
{
|
|
|
|
int g = 0, w = 0, v = 0;
|
|
|
|
|
|
|
|
__asm__ volatile ("mov.u32 %0,%%ctaid.x;" : "=r" (g));
|
|
|
|
__asm__ volatile ("mov.u32 %0,%%tid.y;" : "=r" (w));
|
|
|
|
__asm__ volatile ("mov.u32 %0,%%tid.x;" : "=r" (v));
|
|
|
|
val = (g << 16) | (w << 8) | v;
|
|
|
|
ondev = 1;
|
|
|
|
}
|
|
|
|
t += val;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (ix = 0; ix < N; ix++)
|
|
|
|
{
|
|
|
|
int val = ix;
|
|
|
|
if(ondev)
|
|
|
|
{
|
|
|
|
int g = ix / ((N + 31) / 32);
|
|
|
|
int w = 0;
|
|
|
|
int v = 0;
|
|
|
|
|
|
|
|
val = (g << 16) | (w << 8) | v;
|
|
|
|
}
|
|
|
|
h += val;
|
|
|
|
}
|
|
|
|
if (t != h)
|
|
|
|
{
|
|
|
|
printf ("t=%x expected %x\n", t, h);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|