I tried to compile:
kernel void myufy(int input_position, float4 fuzzy_input[][],float4 vec_ref[][], float3 myu<>)
{
int2 index = instance().xy;
if( fuzzy_input[input_position][index.x].x == vec_ref[index.y][index.x].x )
{
}
else
{
}
}
I know the kernel I built doesn't have output streams, but brcc was compiling without showing when will finish compiling or any issues of error or anything. Windows Task Manager showed brcc active using 1 core in full utilization for a long time until I canceled the operation. After that I revised my code then it works beautifully.
Is it a bug? At first I thought mine had a hardware failure >.<
How to do that?
GPU_coadd_kernel4.domainOffset(uint4(0,0,0,0)); GPU_coadd_kernel4.domainSize(uint4(1,thread_num_coadd,1,1)); GPU_coadd_kernel4(*gpu_temp_coadd_old,*gpu_temp_coadd_working_length,*gpu_temp_coadd);
The problem is, if it can't be compiled, how to define such domain? I'm confused.
Those code to define the domain supposedly to work in host code right? .br resides in another code and another compiler, so I think it might be a brcc's bug.