Hi,

Netscape has a much smaller memory footprint so it comes back faster.
Mozilla with its big memory footprint shows the troubles you will have with 
big memroy systems.
 From my point of view the problem is that the pageoutscanner does not make 
any difference between fs-cache-only-pages and other memory. (until a 
tuneable limit is reached) But the limit does not cope with the problem of 
different workloads.
I have done a small patch that changes the scanner the way that it scans a 
few pages (32) with the lru paradigm and afterwards it scans only for 
fs-cache-only pages (reactivating others) this makes other pages "leaving 
the physical memory" much slower.


Lars

At 09:19 PM 10/25/2001 -0500, Peter Seebach wrote:
>In message <200110260216.f9Q2GCr14080@toad.rmkhome.com>, Rick Kelly writes:
> >The system that I use for chasing -current is a P200 MMX with 128MB. I've
> >been chasing -current on this machine since it went ELF.
>
>Sounds familiar.  I only started again recently, but I mostly keep within
>a couple of days of -current.
>
> >NetBSD roothog.rmkhome.com 1.5Y NetBSD 1.5Y (ROOTHOG) #45: Wed Oct 10 
> 21:57:09
> > MDT 
> 2001     rmk@roothog.rmkhome.com:/usr/src/sys/arch/i386/compile/ROOTHOG i
> >386
>
>Mine's October 15th.
>
> >At the worst point, while using Netscape to read USENET, if I switched to a
> >different virtual desktop (KDE), Netscape would page out completely, and
> >wouldn't always come back.
>
>Mine always comes back.  Eventually.  If I'm doing something large ("tar this
>directory into a pipe"), though, it can take a minute or more.
>
> >I just pulled some iso files over to the -current system. This caused
> >netscape to page out until there was 5 megs left in memory. When the
> >ftp was done I switched to the netscape window and it refreshed in less
> >than 2 seconds.
>
>Yeah.  The problem is definitely worse when I'm doing, say, a "make build"
>while trying to do stuff, but it would be nice to be able to tell the buffer
>cache to be a tad less aggressive.
>
>-s