Welcome, Guest
Username: Password: Remember me

TOPIC: Same case can run with 4 processors but failed with 8 processors

Same case can run with 4 processors but failed with 8 processors 1 year 7 months ago #42317

Hello all,
I am testing on a TOMAWAC case that works with 4 processors,in the 2nd picture, but not with 8 processors, as shown in the 1st picture.

Does anyone encountered this problem and is there any solution?

Many thanks!
Attachments:
The administrator has disabled public write access.

Same case can run with 4 processors but failed with 8 processors 1 year 7 months ago #42326

The 1st picture did not show in the previous post, so I put it in this reply :laugh:
Attachments:
The administrator has disabled public write access.

Same case can run with 4 processors but failed with 8 processors 1 year 7 months ago #42341

  • tfouquet
  • tfouquet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 294
  • Thank you received: 112
Hello,

I guess you do not have enought points to parallelize that much, I think that with 300 points, you don't have to parallelize at all, I guess you do not gain anything, communication between processors makes the gain null when there are so few points per processor.
Probably that the streamline has so few points that the communication between processors is too tricky.

regards.
The administrator has disabled public write access.

Same case can run with 4 processors but failed with 8 processors 1 year 7 months ago #42387

Hello,

Thank you for the advice and it seems that indeed it is the problem that you mentioned. I refined the mesh and the log shows NUMBER OF ELEMENTS: 3833
and NUMBER OF POINTS: 2013, after splitting into 8 parts. Though still it did not work with 8 cores, the calculation could run with 6 cores, which would not work previously.

regard,
The administrator has disabled public write access.
Moderators: tfouquet

The open TELEMAC-MASCARET template for Joomla!2.5, the HTML 4 version.