[stef_pasc]

Hi everybody, I'm trying to validate my code and I discovered that there is a problem in some loops. In particular, I found out that if I make 2 identic loops over cells, I do not obtain the same results. So, in a very simple way, I obtain that the following code gives:

number of cells from loop 1 = 1041
number of cells from loop 2 = 847

The problem is that 2 loops are identical and that I have a mesh of 2024 cells.


DEFINE_ADJUST(adjust_gradient, d)
{
Thread *t;
cell_t c;
int i;

i=0;
thread_loop_c (t,d)
{
begin_c_loop (c,t)
{
i+=1;
printf("boucle cell %d \n", i);
}
end_c_loop (c,t)
}
}


DEFINE_ADJUST(adjust_grad, d)
{
Thread *t;
cell_t c;
int k;

k=0;
thread_loop_c (t,d)
{
begin_c_loop (c,t)
{
k+=1;
printf("boucle cell2 %d \n", k);
}
end_c_loop(c,t)
}
}

Can anybody help me?
Stefano

[blackmask]

You should have got thousands of lines of output. Try to print the result at the end of loop:

DEFINE_ADJUST(adjust_gradient, d)
{
Thread *t;
cell_t c;
int i;

i=0;
thread_loop_c (t,d)
		{
				begin_c_loop (c,t)
				{	
					i+=1;
				}					
				end_c_loop (c,t)
		}
printf("boucle cell %d \n", i);
}

Did you run FLUENT in serial or parallel mode?

[stef_pasc]

Thank you blackman!! Indeed I have thousands of lines but I did intentionally for looking all loops!

I run in parallel mode cause I compile it on Windows. I do not really understand how a so simple action could be wrong!!!

[upeksa]

In not that simple actually, you have to take into account parallel considerations.

Note that you will find very different result according to the number of partitions of your case.

DEFINE_ADJUST(adjust_gradient, d)
{
Thread *t;
cell_t c;
int i;

i=0;
thread_loop_c (t,d)
{
begin_c_loop (c,t)
{
i+=1;
}
end_c_loop (c,t)
}
i=PRF_GISUM1(i);
printf("boucle cell %d \n", i);
}

I think it will work that way, I didn't test it. For further information, read that chapter of your manual.