PDA

View Full Version : Floating point exception in clBuildProgram atiocl64.dll



matrem
03-31-2010, 11:07 AM
With visual c++ 2008 in debug mode, I have a floating point exception (division by zero) in atioocl64... when calling clBuildProgram, does anyone as the same behavior?
In release there is no problem but it's a bit scary.

ps my program :
benchmark.cl

#include "bandwidth.cl"

bandwidth.cl

__kernel

void kernelWrite(__global float * array)
{
array[ get_global_id(0) ] = 123.121f;
}

__kernel
void kernelRead(__global const float * array)
{
float f = array[ get_global_id(0) ];
}

__kernel
void kernelReadWrite(__global float const * array1, __global float * array2)
{
array2[ get_global_id(0) ] = array1[ get_global_id(0) ];
}