[linux-elitists] Performance issues with swap file vs. swap partition

Jason Spence jspence@lightconsulting.com
Thu Nov 6 12:40:20 PST 2003

On Thu, Nov 06, 2003 at 12:05:22PM -0800, Nick Moffitt wrote: 
> begin  Jason Spence  quotation:
> > Huh?  I never noticed there was a problem with IDE in general.  Many
> > times someone will forget to turn DMA on (or cannot due to older
> > hardware) and then I'll see performance problems, but I've never had
> > any issues even with multiple disks going once DMA transfers are
> > enabled.
> 	It's true that you can turn on DMA (note: in SCSI, you
> actually have to find a way to turn it OFF if you're some sort of sick
> person who doesn't like it), but I've got to say that the multiple
> disks thing will kill you.  Even though the drives have DMA, the bus
> gets saturated pretty quick.  You're offloading work from the CPU, but
> the drives are still bone stupid compared to real SCSI peripherals.
> 	I can't comment on SATA, though, other than to note that after
> a while, IDE will have exactly the same feature set (and price point)
> as SCSI-2.

Actually, the SATA II disks debuting early next year have the same
feature set (TCQ, hot-swap, disconnection, etc are mandatory in SATA
II) but are not expected to come out at SCSI price points.  Seagate
especially seems to keep putting out press releases about how cheap
the disks are going to be... these days the only real expensive SATA
disks I can find are the 10k ones like the 36G raptor I have in one of
my boxen.  All the 7200 RPM ones are like $100 or so for 120GB or

After I ask the vendors what's going on, they usually mumble something
about the costs of buying test equipment for their SCSI labs and extra
firmware code, but I haven't had anyone come out and say "here's why
there's this price difference" yet.

 - Jason       Last known location:  11.0 miles northeast of MOUNTAIN VIEW, CA

"I couldn't remember when I had been so disappointed.  Except perhaps
the time I found out that M&Ms really *do* melt in your hand ..."
		-- Peter Oakley

More information about the linux-elitists mailing list