"Robert G. Brown" wrote:
>
> > Intel lacky go home...
>
> Curiously, I actually AM home right now...;-)
>
> I do fail to see why believing fairly passionately that overclocking is
> a fairly stupid thing to do (certainly in a production environment)
> makes me an Intel lackey. I appreciate your irritation at their yanking
> the hardware "standards" world around just as dispassionately as MS
> yanks around the software "standards" world to try to gain advantage
> over their competitors. But overclocking is a different issue
> altogether -- it has been discussed repeatedly on linux-smp and beowulf
> lists and I'd say that the consensus of pros (people whose jobs and work
> depend on a system being up, stable, warranty repairable, and RELIABLY
> GIVING THE RIGHT ANSWERS) is that overclocking is just Russian Roulette
I am at the top of my field and build mission critical machines.
Do I overclock my servers? Generaly no. There is not much point in it. (And the
OC margins are not big enough in the pentium class)
Am I overclocking my workstation? Yes and to an extreme degree with custom
motherboard modifications. http://www.psychosis.com/doa/
Your statement is lacky'ish because it makes an ignorant blanket.
NEVER OVER CLOCK PERIOD. Well shit that is dumb. Just because the sticker says
one thing doesn't mean the part is not designed and capable of something else.
If you know this for sure, they why not use it if it will benifit the
application?
Lemme guess: "Well you can't be sure." Guess what? Neither can Intel. All they
can can say is they tested it at such and such speed, it did not exceed their
maximum tempature spec, and hopefully it will run once you get it.
> -- the question isn't if you'll lose, just when. To give you a very
> BRIEF idea why, my calculations involve a number of operations that
> approaches uncomfortably close to Avogadro's number from below. I'd
> like to get a reproducible answer. I leave it to you just how far I can
> afford to push the statistical probability of failure...anything over
> "zero" is way too big and of course even WITHOUT overclocking it isn't
> quite zero.
Drivel. Moore's law states price/performance doubles every 18 months. Lately
that has been decreasing. This has been my longer stretch inbetween upgrades
ever....22 months. Before pushing the envelope, I was swapping junk every 6-12
months. This time I finally beat Moore out by a long shot.
Who cares if the system dies completly tommorrow. (More realistically in 5
years, instead of 50 : P) How much do you think I can sell a Dual PPro 150 for
now anyhow?
My system is pushed far passed design spec. It could have been questionably
stable. (However it runs like a rock) But taking a PII 300->400 is a no
brainer. It has the same exact core as the faster model, with a production
process that consitently handles 425MHz+. Are you going to tell me the one
tested to run @400MHz by Intel will magically live longer then the one I have
tested? Intels 400 will run without a fan too, right? Phooey. The only point to
not over clocking that is a 'moral' one.
--
http://www.linkscape.net/ Linkscape Internet Services 732-541-4214
http://www.linuxrouter.org/ Linux Router Project
Please, no more BubbaVision.
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
