Subject: Re: UVM/other problems for desktop users in current?
To: NetBSD-current Discussion List <>
From: Greg A. Woods <>
List: current-users
Date: 12/18/2002 03:03:41
[ On Wednesday, December 18, 2002 at 15:32:16 (+1000), George Michaelson wrote: ]
> Subject: Re: UVM/other problems for desktop users in current?
> Ok, within limits I can buy this. But, out of the box, other BSD architectures
> seem to provide a sweeter sweetspot. X may be a dog, but on a Ghz CPU and big
> memory its kinda strange NeBSD can't find the same spot.

Perhaps but multi-GHz CPUs don't necessarily context-switch as fast as
you might think they should, and also remember that unless you've tuned
your VM very carefully for your specific job mix then those I/O
intensive operations are going to push code pages out and there's a good
chance that in a big ugly mess of apps like Xserver + mozilla,
etc. you're going to have a bigger working set of pages than you really
need, especially if you motor your mouse around and wake everything up
and as a result you've got VM competition by those I/O jobs and thus
paging happens and any time you have paging then interactive response
time is going to get spotty.

> Ah, plan9 here we come.

Plan-9 doesn't do anything really new -- it just does what X11+unix
try to do, and does it a whole lot better.

> Yes, but back in the office, in the real world, my
> budget is one box! I'm the tech manager, the last bastion of BSD in a world of
> Linux clones, running a room full of RedHat with NetBSD on me, and no other
> machine. This is *it* its my mail, my X, my filestore, my server, my source
> tree.

Sure, but that just means you have to adjust your expectations a bit and
maybe also tune your system a bit more carefully with an understanding
of what envelope you're trying to push.

> Greg, I really can respect where you are coming from, but this is so ...
> 1970s.

Obviously from your experiences and complaints, it's clearly not.

I don't have any of the problems that have been raised on this thread,
but that's because I separate my GUI and my server.  But then I have
bunches of x-terminals and diskless workstations, and I have a decent
file and compute server, as well as a couple of satellite compute
servers.  When I try to do everything on one box I don't have the same
expectations that I do for my complete network of systems.

Lots of people make sarcastic remarks about how badly so-called "modern"
systems like M$ Windoze-NT make use of way more hardware than any decent
OS needs.  However when you really look at it they're doing what you're
trying to do, and with the same results.  You end up buying ten times
the hardware to do the same job.  I'm happier with my little old NCD
X-terminal and a wimpy old P-Pro-200MHz server with just 64MB RAM than
Peter is with his 2.4HBz 640MB single system.

If you want to run around and look at a whole lot of files all the while
running a bunch of compiles and serving DNS queries and shuffling e-mail
around and serving NFS and FTP and HTTP and so on then you need good and
fast secondary storage system and bunches of RAM and a half-decent CPU
(and maybe a good network card that does some of the grunt work for your
CPU, just like your SCSI disks or FC SAN box are doing).

If you want to push 24-bit pixels around on a display to draw fancy
pictures and make Hi-Fi stereo noises and so on then you need different
kinds of I/O resources and maybe even more RAM to hold all those huge
multi-media applications more CPU power to do all the floating point and
bignum calculations that they require.

If you try and do all of those things on the same box with one lone
not-so-hot-at-context-switching CPU then something's not going to work
so well, especially not if you tune your system to do either alternative
job set even just reasonably well but without thinking about the demands
of the other job set.

> Peter Seebach points out he has a 2.4Gz 640Mb system which is dog slow doing
> CVS. With that much memory/CPU bandwidth, there should be no problem. No?

Yeah, well obviously he hasn't found the right nobs to tweak yet.  :-)

He didn't say what kind of secondary storage subsystem he's got either....

Remember, GHz alone != throughput.  :-)

								Greg A. Woods

+1 416 218-0098;            <>;           <>
Planix, Inc. <>; VE3TCP; Secrets of the Weird <>