Well here is /proc/<pid>/maps


 <<mapxggi>> 
looks like fbdev is included in the map so I assume that the ram on my
Rage128 (16MB) is included in the memory usage. I can't make sense of what
the columns are in this map file so I would appreciate pointers on this one.

Hence XGGI (and ggi etc) would use about 5MB?


Regards,

David

 -----Original Message-----
From:   [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]  On Behalf Of Marcus
Sundberg
Sent:   Wednesday, 10 May 2000 12:01
To:     [EMAIL PROTECTED]
Subject:        Re: XGGI RAM Usage

David Craig <[EMAIL PROTECTED]> writes:

> I am running XGGI/FBDev on PPC and have noticed that its RAM usage is
approx
> 21 MB. This seems rather large and my standard X appears to only use about
> 15MB.

Could you start by checking /proc/<pid>/maps for XGGI and XF86_FBDev
respectively, and see what the RAM is used for (and whether it is RAM
at all, or just framebuffer mem) ? Mail the results to me in any case,
I'd like to have a look as well.

> I compiled using the fullsrc cut down version and apart from reducing my
> Fbdev resolution I do not know how I might get it more lean. (the
hosts.def
> file already seems to be preconfigured for minimal support).
> 
> I only need basic X support on fbdev only running a couple of windows and
> apps, can anyone give me some pointers on how I might be able to optimise
> XGGI further?

One way is to skip out the cfb renderers for color depths you'll
never use, but I have never looked at this myself. I really find
it strange that XGGI should use so much more RAM, so before saying
anything else I'd like to check those maps files.

Btw, I tried to test your XGGI binary, but it seems to be corrupted
in some way. Trying to run it causes an instant segfault in the
dynamic linker, and trying to run any of my cross-binutils
(objdump/nm/strings) on it crashes them because they are trying to
malloc() 1.5GB of memory... Could you gzip it before sending so
there'll be some checksumming.

//Marcus
-- 
-------------------------------+------------------------------------
        Marcus Sundberg        | http://www.stacken.kth.se/~mackan
 Royal Institute of Technology |       Phone: +46 707 295404
       Stockholm, Sweden       |   E-Mail: [EMAIL PROTECTED]

mapxggi

Reply via email to