Subject: Re: UVM/other problems for desktop users in current?
To: None <firstname.lastname@example.org>
From: Wolfgang Rupprecht <wolfgang+gnus20021217T212800@wsrcc.com>
Date: 12/17/2002 21:31:34
email@example.com (Greg A. Woods) writes:
> Sure you can get away with running Xserver and one big application at a
> time like Mozilla or what have you on a workstation, but that's because
> you're essentially doing primarily GUI stuff. However once you start
> compiling and running I/O intensive stuff like CVS then you're doing
> more what's done on a multi-user general-purpose server. Those two
> kinds of application just do not mix well on a single CPU system -- you
> can't expect to get good compiler throughput and at the same time have
> sub-second GUI response, especially when you context switch through a
> half dozen big GUI applications all at once too!
Perhaps one has to break the loads down into at least two major
categories. 1) CPU saturating loads 2) disk-arm saturating loads.
I've seen no problem with running cpu-intensive tasks such as "nice
-20" userland builds while XFree86 was running. On the other hand,
things do get slow when someone is saturating the disk arm.
I've always had a sneaking suspicion that the disk arm could use a
better scheduler. I bet it would make things feel quite a bit
snappier if all the interactive tasks got first dibs on the disk arm
and the non-interactive things (like 'cvs update') got last dibs.
Some niceness-like value that trickled down into the disk buffers
could allow the sweep algorithm to favor buffers that came from
Wolfgang Rupprecht http://www.wsrcc.com/wolfgang/
spider food: http://www.wsrcc.com/baddream/usenet/
(NOTE: The email address above is valid. Edit it at your own peril.)