[info-performer] Re: help with pfMemory and PFSHAREDSIZE

Date view Thread view Subject view Author view

From: Steve Satterfield (steve++at++nist.gov)
Date: 01/09/2005 20:01:20

Thanks to Jimmy Wang of SGI, I got a hint of why my application runs on an
Octane2 and Linux but not on Onyx4. It seems to be related to multi processing
on the larger machines.

My application uses perfly and a locally developed file loader.

This command will run on Octane2 but not on Onyx4:

    setenv PFSHAREDSIZE 1000000000
    setenv PFSHAREDBASE 0x20000000
    perfly demo.seq

However, this command will run on the Onyx4

    perfly -m 0 demo.seq

So now my question is why does multiprocessing reduce the apparent size of the
shared memory? Do the various stages of Performer copy parts or all of the
scene graph? Is there anything I can to other than restrict the Onxy4 to a
single process?

Thanks,
Steve

On Fri, 7 Jan 2005, Steve Satterfield wrote:

> Hi,
>
> I am trying to run a N32 Performer application on an Onyx4 but it fails with
> the pfMemory error:
>
> PF Fatal/Resource: pfMemory::new() Unable to allocate 59576 bytes from arena 0x20000000.
> Try using pfSharedArenaSize() or env PFSHAREDSIZE
> to increase the arena size
> (currently 976562.50 KBytes) and check for adequate setrlimit()
> values and available space on swap (or pfTmpDir()).
>
> When I run this application, I have
> PFSHAREDSIZE set to 1000000000
> PFSHAREDBASE set to 0x20000000
>
> The Onyx4 is 4 CPU/4GB memory running 6.5.24 and OpenGL Performer 3.1.1
>
> This same application will however run well on an Octane2 with 2 CPU/512Mbytes
> memory running 6.5.24 and OpenGL Performer 3.1.
>
> Additionally, this application also fails on an Onyx3 with 24CPU/24GB memory
> running 6.5.24 and OpenGL Performer 3.1.1. It also fails on a different 8
> CPU/8GB memory Onyx4.
>
> Since it runs on a desktop machine but not on larger machines, I assume there
> must be something configured differently about the larger machines.
>
> Can you give me any ideas on where to look for differences in the larger
> machines?
>
> I have compared the values of "systune shm". The larger machines have larger
> values. As a test, I changed the Onyx4 systune shm parameters to match the
> Octane systune shm values, but the application continues to fail.
>
> Thank you for any help you can provide.
>
> -Steve
>
> ----------------------------------------------
> Steven G. Satterfield steve++at++nist.gov
> NIST (voice) 301 975-5637
> 100 Bureau Dr, Stop 8911 (fax) 301 975-3218
> Gaithersburg, MD 20899-8911
>
>

Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Sun Jan 09 2005 - 20:01:31 PST