CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > ANSYS > FLUENT

Parallel fluent not using all processors specified

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   March 31, 2008, 15:18
Default Parallel fluent not using all processors specified
  #1
Paul
Guest
 
Posts: n/a
I'm running an 8-xeon processor workstation on xp-64. I have two instances of fluent running right now, each are -t3. Why is it that if I go to the windows tast manager and look at the CPU usage history it appears that only 6 of the processors are being used? The CPU usage for the computer overall is at 75%, but shouldn't it be 100% since each instance is using four processors?

I'd be rather annoyed if all these processors and licensing fees are not being put to their full use.

Thanks.
  Reply With Quote

Old   March 31, 2008, 15:48
Default Re: Parallel fluent not using all processors speci
  #2
Remus
Guest
 
Posts: n/a
It looks that you asked each Fluent instance to run on 3 processors according to -t3 option. That gives a total of 6 out of 8 processors which is 75% overall. I don't see any error.

Remus
  Reply With Quote

Old   March 31, 2008, 16:21
Default Re: Parallel fluent not using all processors speci
  #3
John
Guest
 
Posts: n/a
I'm pretty sure that -t3 means three additional processors to the first one, so four total for each.
  Reply With Quote

Old   March 31, 2008, 16:29
Default Re: Parallel fluent not using all processors speci
  #4
Remus
Guest
 
Posts: n/a
Not in my case. That is the total number of CPUs requested (at least for Linux). Remus
  Reply With Quote

Old   March 31, 2008, 16:29
Default Re: Parallel fluent not using all processors speci
  #5
Paul
Guest
 
Posts: n/a
Thats what I though, or at least thats what my boss told me. Something is definitely odd about it though. The license should be for eight processors on this machine.
  Reply With Quote

Old   March 31, 2008, 16:43
Default Re: Parallel fluent not using all processors speci
  #6
Paul
Guest
 
Posts: n/a
Okay when I load up fluent 3d -t4 the console gives me this:

------------------------------------------------------------------------------ ID Comm. Hostname O.S. PID Mach ID HW ID Name ------------------------------------------------------------------------------ host net Sys156 Windows-x64 2060 0 1780 Fluent Host n3 mpich2 Sys156 Windows-x64 2708 0 3 Fluent Node n2 mpich2 Sys156 Windows-x64 3068 0 2 Fluent Node n1 mpich2 Sys156 Windows-x64 1272 0 1 Fluent Node n0* mpich2 Sys156 Windows-x64 3608 0 0 Fluent Node

------------------------------------------------------------------------------

It says it is using four nodes and one host. Is this normal? It seems to be taking the four processors, and then processor #5 is running as the host.

As an aside, I cannot load any higher than -t2 at this point because fluent won't let me (again the license is only for eight total processors).
  Reply With Quote

Old   March 31, 2008, 17:04
Default Re: Parallel fluent not using all processors speci
  #7
Paul
Guest
 
Posts: n/a
I've looked into it some more and it appears that the first processor is designated "the host", and does not assist in the solving of the simulation.

I don't suppose theres any way to make it act as a node is there?
  Reply With Quote

Old   March 31, 2008, 17:15
Default Re: Parallel fluent not using all processors speci
  #8
Remus
Guest
 
Posts: n/a
When I type fluent 3d -t4 what I get is a host process and 4 nodes (processors) allocated. So, if I want, I can start another instance of parallel Fluent with the same command line (fluent 3d -t4) and it's going to work fine since I'll be using 8 processors. As a test, what happens if you try to start fluent on 8 processors (fluent 3d -t8)?

  Reply With Quote

Old   March 31, 2008, 17:18
Default Re: Parallel fluent not using all processors speci
  #9
Paul
Guest
 
Posts: n/a
-t8 gives me errors telling me that all possible instances of fluent are in use. The host process appears to make use of its own processor, and that processor is not part of the grid. I checked the auto-partitioning and it is indeed only partitioning for 3 nodes because that is all it reads.

Remus, are you saying that you have a similar eight processor system, but are still able to run two processes with four nodes each (and two hosts)? Is your license supposed to be maxed out at eight processors?
  Reply With Quote

Old   March 31, 2008, 17:56
Default Re: Parallel fluent not using all processors speci
  #10
Remus
Guest
 
Posts: n/a
I have a 12 processor license and I can run any combination of two instances with -tx and -ty as long as x+y=12. How is your license structured, i.e., how many serial and how many additional parallel licenses?
  Reply With Quote

Old   March 31, 2008, 18:05
Default Re: Parallel fluent not using all processors speci
  #11
Paul
Guest
 
Posts: n/a
I believe its 2 serial (meaning I can run two instances total of fluent and/or gambit but no more), and then six parallel processors on top of that (at least it should be).

I'll submit a support ticket to fluent and see what they say.
  Reply With Quote

Old   March 31, 2008, 18:20
Default Re: Parallel fluent not using all processors speci
  #12
Remus
Guest
 
Posts: n/a
Correction: x+y <= 12
  Reply With Quote

Old   April 2, 2008, 14:22
Default Re: Parallel fluent not using all processors speci
  #13
Paul
Guest
 
Posts: n/a
Got a response:

"The host process takes up one of the processes in your parallel license. I guess if you had 9 processes in your license you could have had 8 computing nodes and one of the nodes would have been shared with the host."

What a load of shit.
  Reply With Quote

Old   April 2, 2008, 16:45
Default Re: Parallel fluent not using all processors speci
  #14
R.
Guest
 
Posts: n/a
I also have an 8 core server. I have 8 active and 10 parallel lics

When I run fluent (linux environment) as 'fluent -t8', 8 nodes are spawned to compute the solution along with one host, and one active lic and 7 parallel lics are checked out.

fluent -r6.3.26 3d -node -alnx86 -t8 -pethernet -mpi=hp Starting

------------------------------------------------------------------------------ ID Comm. Hostname O.S. PID Mach ID HW ID Name ------------------------------------------------------------------------------ host net shelob Linux-64 20436 0 3 Fluent Host n7 hp shelob Linux-64 20534 0 7 Fluent Node n6 hp shelob Linux-64 20533 0 6 Fluent Node n5 hp shelob Linux-64 20532 0 5 Fluent Node n4 hp shelob Linux-64 20531 0 4 Fluent Node n3 hp shelob Linux-64 20530 0 3 Fluent Node n2 hp shelob Linux-64 20529 0 2 Fluent Node n1 hp shelob Linux-64 20528 0 1 Fluent Node n0* hp shelob Linux-64 20527 0 0 Fluent Node

Selected interconnect: ethernet (intra-machine comm. may use shared memory) ------------------------------------------------------------------------------

Heres something you may find useful:

Product Fluent6 Version No. 6.3.26

Problem:

A 4-process parallel FLUENT session runs on a machine equipped with 4 dual-core opteron processors. A second session of the same kind is launched: It uses the same 4 processor cores, leaving the other 4 processor cores unused (idle). (This does not necessarily happen on all 8-cpu/core machines, but it has been observed on some.)

Resolution:

In one of the session, type the following into the FLUENT text console window: (set-affinity "0 4 1 5 2 6 3 7") This will move the node processes 0--3 to the processor cores 3--7. The strings contains a sequence of pairs of numbers, each pair consisting of a process id and a cpu/core id. Process ids are 0, 1, 2,... for the node processes and 999999 for the host process.

  Reply With Quote

Old   April 3, 2008, 09:16
Default Re: Parallel fluent not using all processors speci
  #15
Paul
Guest
 
Posts: n/a
Yeah I've seen that support case, but its really not applicable to me.
  Reply With Quote

Old   April 3, 2008, 15:10
Default Re: Parallel fluent not using all processors speci
  #16
rj
Guest
 
Posts: n/a
This seems strange response. FLUENT can use either FLUENT licenses or ANSYS licenses.

1. If use FLUENT licenses, the used license number should be equal to the number of node processes, while the host process will not occupy any license. You should be able to run -t8 if you have 8 licenses.

1. If use ANSYS licenses, the used license number should be equal the number of nodes plus 1, i.e., the host process will occupy a license. You should be able to run -t7 if you have 8 licenses.

Hope this helps to explain.
  Reply With Quote

Old   April 7, 2008, 09:19
Default Re: Parallel fluent not using all processors speci
  #17
Aris
Guest
 
Posts: n/a
I have fluent licence for two processings and 8 parallel.

I can run fluent -t8. One host-8 nodes. All 8 cores of my system work. I can also run 2 cases with -t4. Because i have licence for two processings.

In your case i believe you can run 1 case with -t6. Or two cases with -t3. In any case you cannot use the 2 out of 8 cores you have. You should upgrade your license.

  Reply With Quote

Old   April 7, 2008, 14:18
Default Re: Parallel fluent not using all processors speci
  #18
Paul
Guest
 
Posts: n/a
I can one one case with -t7, or two with -t3 each.

And upgrading the license isn't an option right now. This shit costs enough as it is.
  Reply With Quote

Old   October 26, 2023, 04:54
Default
  #19
Member
 
Join Date: Oct 2022
Posts: 61
Rep Power: 3
@Arash is on a distinguished road
Quote:
Originally Posted by R.
;149619
I also have an 8 core server. I have 8 active and 10 parallel lics

When I run fluent (linux environment) as 'fluent -t8', 8 nodes are spawned to compute the solution along with one host, and one active lic and 7 parallel lics are checked out.

fluent -r6.3.26 3d -node -alnx86 -t8 -pethernet -mpi=hp Starting

------------------------------------------------------------------------------ ID Comm. Hostname O.S. PID Mach ID HW ID Name ------------------------------------------------------------------------------ host net shelob Linux-64 20436 0 3 Fluent Host n7 hp shelob Linux-64 20534 0 7 Fluent Node n6 hp shelob Linux-64 20533 0 6 Fluent Node n5 hp shelob Linux-64 20532 0 5 Fluent Node n4 hp shelob Linux-64 20531 0 4 Fluent Node n3 hp shelob Linux-64 20530 0 3 Fluent Node n2 hp shelob Linux-64 20529 0 2 Fluent Node n1 hp shelob Linux-64 20528 0 1 Fluent Node n0* hp shelob Linux-64 20527 0 0 Fluent Node

Selected interconnect: ethernet (intra-machine comm. may use shared memory) ------------------------------------------------------------------------------

Heres something you may find useful:

Product Fluent6 Version No. 6.3.26

Problem:

A 4-process parallel FLUENT session runs on a machine equipped with 4 dual-core opteron processors. A second session of the same kind is launched: It uses the same 4 processor cores, leaving the other 4 processor cores unused (idle). (This does not necessarily happen on all 8-cpu/core machines, but it has been observed on some.)

Resolution:

In one of the session, type the following into the FLUENT text console window: (set-affinity "0 4 1 5 2 6 3 7") This will move the node processes 0--3 to the processor cores 3--7. The strings contains a sequence of pairs of numbers, each pair consisting of a process id and a cpu/core id. Process ids are 0, 1, 2,... for the node processes and 999999 for the host process.

Hi Dear friend,

I use an Intel core i9-13900k processor that has 24 physical cores. When I use the number of cores for example 14 or 16 or... in order to parallelize and reduce the solution time, my solution goes slowly. What should I do ?
I selected 16 cores in the Fluent window and then imported the mesh file into Fluent. As soon as Fluent opens in the console, I receive the following message:


Note: Rank = 0: Process affinity not being set (6).
Note: Rank = 1: Process affinity not being set (6).
Note: Rank = 2: Process affinity not being set (6).
Note: Rank = 3: Process affinity not being set (6).
Note: Rank = 4: Process affinity not being set (6).
Note: Rank = 5: Process affinity not being set (6).
Note: Rank = 6: Process affinity not being set (6).
Note: Rank = 7: Process affinity not being set (6).
Note: Rank = 8: Process affinity not being set (6).
Note: Rank = 9: Process affinity not being set (6).
Note: Rank = 10: Process affinity not being set (6).
Note: Rank = 11: Process affinity not being set (6).
Note: Rank = 12: Process affinity not being set (6).
Note: Rank = 13: Process affinity not being set (6).
Note: Rank = 14: Process affinity not being set (6).
Note: Rank = 15: Process affinity not being set (6).


When using the number of cores, for example 14 or 16 or... my solution is slow and very annoying.
Please help me what to do?

Thankyou
@Arash is offline   Reply With Quote

Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Find neighbours across processors in parallel computing xiao OpenFOAM Running, Solving & CFD 1 December 15, 2008 13:14
Parallel with 2 processors dual-core. Sam FLUENT 4 October 10, 2007 11:43
Parallel Computing on Multi-Core Processors Upgrading Hardware CFX 6 June 7, 2007 16:54
Does Fluent runs better on 4 processors than 2? Yi FLUENT 1 October 28, 2005 21:02
Parallel processors Joel Phoenics 1 April 18, 2002 09:09


All times are GMT -4. The time now is 06:36.