From: Allan Schaffer (allan++at++sgi.com)
Date: 01/02/2004 15:35:23
Hello Anbu, happy new year --
The general problem is that Performer's shared arena must be a contiguous
block of memory; but various system libraries (mostly Performer's) are
sprinkled around in memory to the point where it's difficult to get a very
large region.
The good news: we've fixed this in the final Performer 3.1 images. These
will be available in days.
In the meantime:
- Besides PFSHAREDSIZE you can also experiment with different values for
PFSHAREDBASE to get a lower starting point than the default.
Otherwise you (or we) need to remap your Performer libraries into different
locations in memory, hopefully more tightly packed together.
- We ship a tool /usr/share/Performer/src/tools/remapLibs which can do this
for N32 and O32 applications. It maps them into high memory.
- If you are an rqs expert: remap your N32 libraries to this location:
:st = $range 0x02000000, 0x08000000:
N64 libraries:
:st = $range 0x02000000, 0x0b980000:
That's relatively low memory. This is what we've done for 3.1.
Allan
Anbumani wrote:
> We have problem in loading pfb data ( 450MB) in the Onyx (3200 Infinite
> reality) system.
>
> our system config:
>
> CPU : 4 nos of 500Mhz Mips R14000
> Main memory :8 GB
> OS : Irix64 release 6.5
>
> When i try to load 450mb, it is giving the following error message:
>
> pfMemory::new() Unable to allocate 59412 bytes from
> arena 0x49590000. Try using pfSharedArenaSize() or env
> PFSHAREDSIZE to increase the arena size (currently
> 65536.00 kBytes) and check for adequate setrlimit()
> values and available space on swap (or pfTmpDir()).
>
> we can load upto 380mb.
> We tried to increase shared arena size as specified above.
> but still the problem exists.
> Pls, help me to resolve the problem.
-- Allan Schaffer allan++at++sgi.com Engr. Manager, Core Rendering 1-650-933-2160 Silicon Graphics http://www.sgi.com
This archive was generated by hypermail 2b29 : Fri Jan 02 2004 - 16:42:58 PST