[xh110120]

My simulation is transient for the calculation of pressure wave in the domain. And I want to know the maximum pressure and minimum pressure of the domain after every timestep. So I use the "DEFINE_EXECUTE_AT_END" to see the value, the program can be seen in the picture below. I use MAX(pmaxx,C_P(c,t)) and MIN(pminx,C_P(c,t)) to get the maximum pressure and minimum pressure for every partitions(28 partitions for the parallel calculation). And I use PRF_GRHIGH1(pmaxx) and PRF_GRLOW1(pminx) to get the maximum and minimum pressure for all of the domain. In theory, the value of pmax and pmin should be the same for every partition. But in the console, the first partition is different from the other partitions, and the other partitions are the same. The result can be seen in the picture below. And I wonder why the first partition is different, which will affect my following calculation.
Thank you for your help.
program.jpg

result1.jpg

result2.jpg

result3.jpg

[pakk]

It is not the first partition that is different, but the first process, the host process. The host process does not deal with geometry, so you should exclude that.

For example, by changing the writing-part of your code to:

#if RP_NODE
 Message("Max:%e;Min:%e\n",pmax,pmin);
#endif /* RP_NODE */

In this way, only the results for compute nodes will be shown.

[xh110120]

Originally Posted by pakk

It is not the first partition that is different, but the first process, the host process. The host process does not deal with geometry, so you should exclude that.

For example, by changing the writing-part of your code to:

#if RP_NODE
 Message("Max:%e;Min:%e\n",pmax,pmin);
#endif /* RP_NODE */

In this way, only the results for compute nodes will be shown.

Hey pakk,
Your answer is helpful. And I want to use the pmin and pmax to do the logical judgment for the next calculation.(The detail program is below) And I wonder whether it use the pmax and pmin in the host process or in the node process to do the logical judgment. I want it use the value in the node process to do this work, because the value in the node process is the real maximum and minimum pressure in this domain. And what can I do to realize it ?
Thank you very much!

thread_loop_c(t,d)
{
begin_c_loop_int(c,t)
{
if(C_P(c,t)>=(pmin+ratio*(pmax-pmin)))
C_UDMI(c,t,2)=2;
if((pmin+ratio*(pmax-pmin))>C_P(c,t)>=pmin)
C_UDMI(c,t,2)=1;
if(C_UDMI(c,t,0)!=1&&C_UDMI(c,t,2)==1)
x+=(C_VOLUME(c,t)*2*PI);
if(C_UDMI(c,t,0)!=1&&C_UDMI(c,t,2)==2)
x+=(k*C_VOLUME(c,t)*2*PI);
}
end_c_loop_int(c,t)
}

[pakk]

Probably you can achieve what you want by putting "#if RP_NODE" and "#endif" around your code you showed here.

I would advise you to read section 7 (Parallel considerations) of the UDF manual. It has some clear examples of this.

[xh110120]

Dear pakk,
Thank you for your help. By arounding my code with the "#if RP_NODE" and "#endif", the problem indeed work out!