|
[Sponsors] |
March 1, 2007, 11:43 |
over
|
#1 |
Guest
Posts: n/a
|
hi,
I am conducting a compressible CFX10 simulation. The weired thing is, the simulation runs ok on the local PC, but always fails on linux cluster. The error information is overflow. How to fix it? Thanks a lot Equation | Rate | RMS Res | Max Res | Linear Solution | U-Mom | 0.00 | 1.3E-12 | 3.7E-11 | 4.7E+08 F | V-Mom | 0.00 | 9.2E-12 | 1.5E-10 | 2.4E+08 F | W-Mom | 0.00 | 3.6E-12 | 7.2E-11 | 6.3E+08 F | P-Mass| 0.00 | 4.2E-22 | 2.0E-20 | 9.3 4.5E+08 F | ERROR #001100279 has occurred in subroutine c_fpx_handler: Floating point exception: Overflow |
|
March 1, 2007, 11:59 |
Re: over
|
#2 |
Guest
Posts: n/a
|
Try running benchmark.def with two partitions on the cluster to eliminate it as the source of the problem.
|
|
March 1, 2007, 14:13 |
Re: over
|
#3 |
Guest
Posts: n/a
|
hi, thanks! Can you explain more in detail how to fix it? Also, we don't have parallel simulation license on either pc or cluster.
|
|
March 1, 2007, 14:28 |
Re: over
|
#4 |
Guest
Posts: n/a
|
Test benchmark.def serially on both your windows machine and linux cluster to get a baseline going.
|
|
March 1, 2007, 14:36 |
Re: over
|
#5 |
Guest
Posts: n/a
|
what is benchmark.def? A test case simulation? Previously, our cluster always ran very well with CFX10. This is the first time I have this trouble.
|
|
March 2, 2007, 01:21 |
Re: over
|
#6 |
Guest
Posts: n/a
|
Check the /examples directory. It's there in CFX 11.0, not sure about 10.0, but assume so.
|
|
Thread Tools | Search this Thread |
Display Modes | |
|
|