On Sat, Jun 19, 1999 at 04:57:25PM -0600, Warner Losh wrote:
> By broke, I mean that they don't scale well due to cache effects. I
> think it may just be a size thing, but it might also be a cache
> coherency protocol ineffeciencies as well. The articles I've seen
> show that for typical workloads, people with two celerons were getting
> in the 1.5x range, while people with PIIs were getting 1.8x or so.
>
> Warnr
Warner,
Don't suppose you have a URL handy for those articles? I'm contemplating
building a dual-Celery box so it's probably good to know this stuff.
Although, with the Celerons over here selling for around half the price of
an equivalent P-II they'd have to scale real bad to put me off the idea :-)
Even if the performance sucks, I'll have an otherwise well-specced box that
I can upgrade to P-II's as I can afford it.
Cheers,
Scott
--
===========================================================================
Scott Mitchell | PGP Key ID | "Eagles may soar, but weasels
London, England | 0x54B171B9 | don't get sucked into jet engines"
s.mitch...@computer.org | 0xAA775B8B | -- Anon
To Unsubscribe: send mail to majord...@freebsd.org
with "unsubscribe freebsd-hackers" in the body of the message
