Subject: Re: Very slow disk.
To: Jeff Northon <jeffo@sasquatch.com>
From: Richard Rauch <rauch@rice.edu>
List: netbsd-help
Date: 12/30/2001 01:24:02
On Sat, 29 Dec 2001, Jeff Northon wrote:

> http://www.theregister.co.uk/content/3/23502.html

Interesting, but if I could get 64 to 90 MB/sec from my system, I'd be
quite happy.  I'm also not doing any form of RAID, nor is my disk a SCSI.

(In fact, comparing bonnie++ output between this and another machine, the
only ``benchmark'' rates that are slower are sequetial read and sequential
delete.  Everything else is much faster (as much as 4 or 5 times as fast)
with the VIA/Athlon combination.)

(Of course, I know not to place complete trust in benchmarks, and what
bonnie++ calls ``sequential read'' could be exactly what I'm getting
clobbered over...)


> http://pcbuyersguide.com/hardware/motherboards/VIA-Problems.html

Interesting reading...though other than a reference back to the
theregister page, there's nothing helpful for my situation on it.  This
seems to be a general ``we don't like VIA'' page.  (Well, if I were
running MS-WINDOWS, with an AGP graphics card, etc., I might find some
more information useful...)


> http://www.tweakers.net/meuktracker/628

Regrettably, I can't read German, so I can only get the general gist
(``we don't like VIA'' seems to come through again).


 [...]
> I saw there was a few VIA BIOS flash upgrades available for the AMD / VIA
> and tried the newest (which helped) but did not solve all the slowness /
> data corruption.

Presmably, I'd need an MS-DOS/MS-WINDOWS installation in order to use the
upgrades?  I seem to recall that these flash upgrades come in the form of
.EXE files for MS-DOS or MS-WINDOWS.  I'm not sure how I would go about
running them on my systems.  (^&

In any case, I have *no* observed data corruption problems.  If
``upgrading'' would risk introducing data corruption problems, I'd rather
have the slower (but working) stuff in there.  (Aside: Why should BIOS
affect performance of the chipset?  NetBSD doesn't use BIOS once NetBSD is
booted, does it?)


> I do like AMD and wish they could tighten the screws on chipset companies
> to avoid these types of problems. (They probably can't twist arms or fund
> development the way Intel does).

Probably true.

Still, one would think that there would be a market for a solid chipset
that worked with AMD CPU's.


Oh well.  Thanks for the HTTP: pointers.  I'm not sure how much it really
reflects my present system, but it's definitely something to bear in mind
the next time that I'm in the market for a new system.  (Unless I finally
dig myself back out of the i386 market.)


  ``I probably don't know what I'm talking about.'' --rauch@math.rice.edu