On Sun, 21 Nov 1999, Dan Swartzendruber wrote:
> At 01:05 PM 11/21/99 -0700, Axalon Bloodstone wrote:
> >On Sun, 21 Nov 1999, Steve Philp wrote:
> >[...]
> >>
> >> Overall, though, I didn't notice enough of a speed difference between
> >> the drive on UDMA/66 and the drive on a "normal" interface to want to
> >> fight with it anymore. I'm perfectly happy with the drive on a normal
> >> interface.
> >
> >With it on,
> >
> >/dev/hda:
> > Timing buffer-cache reads: 128 MB in 1.75 seconds =73.14 MB/sec
> > Timing buffered disk reads: 64 MB in 21.91 seconds = 2.92 MB/sec
> >
> >I'm not gonna switch it just to show you guys, but on hde cache reads hit
> >around 125-150, and buffered gets up to about 5-10. A mighty big
> >improvment, but....
>
> Well, yeah, but this isn't a real meaningful test. A real-world app is
> still going to be limited primarily by how fast it can spin the platters...
;) all tests are meaningless. Notice i didn't mention system load or any
other specs. It could have spit out "128 MB in 0.01 seconds" and it's
still not worth my time to d**k around with (ok maybe if it really said
that but it doesn't)
The point was it's not worth it _to me_ (and apparently Steve, as much as
he'd like to see it work)
