Re: adding SCSI C: to IDE system


Rod Speed wrote:
> Cruise Director <SeaFuncSpam@xxxxxxxxx> wrote
>
> > Are you so experienced and so well informed that you've
> > been all over the academic performance literature?
>
> Dont need to be. If the effect was anything like you are claiming, you'd
> see plenty of recommendations to buy a new drive when the current
> drive gets over 50% full. You dont, so you're remembering wrong.

On the humorous side, this reminds me of the joke about 2 economists.
2 economists are walking down the street. A $20 bill flaps in the
wind, stuck near a gutter. 1 economist says to the other, "Look, a $20
bill!" The other says, "That's impossible. If it were, someone would
have picked it up already." And they keep right on walking.

On the serious side, nobody who does benchmarking professionally
believes in benchmarking theory. They believe in empirical results.
Whether you can ascribe a rationale for such a phenomenon is
irrelevant. I have not benchmarked hard drives professionally, but I
have done it for 3D graphics cards, and I know there are *many* things
that affect performance. Many that I would rather not have exist,
because they usually demonstrate what a shithole the PC industry is.

Anyways I've followed up to comp.benchmarks and that'll be the endpoint
of my quest on the issue. Either someone concurs, someone disagrees,
or - the most likely - crickets chirp.

> And you can prove it for yourself any time you
> like too. Its a completely trivial claim to test.

Well it would take a little shell scripting I suppose. And also some
massaging of results to make nice charts. I would call this an "easy"
project, not a trivial one. I tend to measure "trivial" in terms of
"can I do it in 1 hour?" Sounds more like it would take a day. I've
got better things to spend a day on right now, ergo a post to
comp.benchmarks.

> >> Dont believe it. And you dont see recommendations
> >> to upgrade to bigger drives when they are half full,
> >> and you would if it was a real effect.
>
> > I'm not convinced the marketing depts. would play it as you say.
>
> Nothing to do with marketing depts, everything to do with what
> operations like storagereview, Tom's, PC Mag etc do all the time.

PC Mag is most certainly a marketing dept. They only piss off their
major advertizers in exceptional circumstances, so that they can
maintain an air of credibility. StorageReview and Tom's are more
honest. Actually it's a good forum question for them.

> > It is possible the researchers themselves
> > didn't choose to separate effects.
>
> Or you are remembering it wrong.

Yes, either is possible. However, I think it likely my memory is
intact on this one.


Cheers,
Brandon Van Every

.