News Focus
News Focus
icon url

Elmer Phud

03/14/03 12:01 PM

#959 RE: yourbankruptcy #956

Yourbankruptcy -

If you claim that new 50% faster Itanium will come soon, then don't forget that Opteron scales better and is smaller and cheaper

While we're busy reminding each other, let me remind you that you have no data to support your claim about Opteron. You have only promises from a company that has yet to deliver anything. For Intel to claim Madison will increase performance by 50% by essentially moving to a proven .13u process and adding more cache is a much more believable promise from a company who has already demonstrated the capability.
icon url

wbmw

03/14/03 12:24 PM

#960 RE: yourbankruptcy #956

YB, Re: If you think 4-way Madison will be faster than 4-way Opteron (which is not a fact, maybe it will not), just take 8-way Opteron. You don't have to compare 4-to-4, why should you? Compare rack-to-rack.

An interesting premise, but not every company buys server systems with rack density in mind. Racks are more of a front-end convenience, but Itanium is being aimed at the back-end. Even if there are 8-way Opteron systems this year (which I highly doubt, given that no one has announced any plans for this), it will likely be compared against other 8-way Itanium 2 systems. The reason is that besides additional performance, larger CPU scaled systems carry additional features. These include more I/O, more RASUM, and more memory capacity. Opteron can certainly support these things, but the platform still needs to be designed to accommodate them. That means more I/O backplanes, more memory add-in boards, more server management hooks, etc. This is much more than any whitebox provider can offer. Unless AMD gets a major tier-1 or tier-2 design win with an investment to create the infrastructure for a large scaled-up Opteron platform, I wouldn't expect to see any.