On Tuesday 25 May 2004 11:36 am, Derrick Ryalls wrote:
> <snip>
> > FWIW, none of my new systems have floppies and the smallest
> > has 3 x 40GB HDs. The new HDs all run ATA-133 and have 8MB of
> > cache. I have about 40GB of mp3 and wma that I have created
> > from CDs that I own and backup on one of the 2400's. It isn't
> > the one with the 3-40's :). None of the on-board audio
> > connect to the CD-Rom with a digital connection. Digital
> > extraction turns out to be very important when converting audio
> > cds.
> >
> > An ATA-133 controller is also important because you will find
> > it difficult to buy HDs under 120GB and the older mobo may
> > not recognize the newer, large ones. A 2400+ with ATA-133 HDs
> > will do a buildworld in
> > 18 minutes.
> That makes me wonder what I am doing wrong.  I have an 2500XP, gig of
> ram, and a SATA drive, and buildworld take about 45min, buildkernel
> about 5-10min.  Yes, timing measured from pure console, not X.

I have /usr, /usr/src, and /usr/obj on their own ATA-133 controllers. 
When I was benchmarking Cray's, the kind of write caching you were 
doing could double the ouput. You wanted the write to be allowed to 
wait so that reads had priority. If you are running a corp. data base, 
data integrity if more important but this is my own machine and I 
wanted turn around speed on the buildworld.

A buildworld on 5.x is about 50% slower than 4-stable. On a single cpu, 
using -j? actually slowed the build down. I do not use the -j option.


Kent Stewart
Richland, WA

[EMAIL PROTECTED] mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to