The Inquirer-Home

Intel performance suffers frequency hit again

Dempsey is the latest victim
Fri Nov 18 2005, 07:42
ANYONE NOTICE THE LITTLE oopsie in the latest Intel Xeon Roadmap?

Look at the table on the bottom, and notice that the Dempsey chips numbered 5020/30/40/50/60/70 are either at a 1066MHz FSB or a 667 FSB. Curious that.

Now, Dempsey has none of the problems that the old Pentium 4s had at hitting this bus speeds, and from what we gather, they are yielding just fine at that speed. The problem as we understand it is the PLLs (phase locked loops) are still carried over from the old P4s, and they can't go below a 12x multiplier. Intel was supposed to be way beyond this speed, pushing 5GHz by now instead of pining for 3.5GHz, so it wasn't meant to be a problem. So, below 12x or 3.0GHz, you need to drop the bus down by a notch.

Now, you would think that they would go from 1066 to 800, right? That would make sense but the problem is that Blackford won't support an 800 front side bus (FSB), hence the oopsie. Now, we would put this in the category of boneheaded design choice, but would take most of it back because it was designing the chipset for a much different CPU. Had the CPU side delivered, this never would have arisen, but now it is going to hit performance hard.

What this means is that the 3.2GHz and 3.46GHz Dempsey/Blackford combos are going to do quite well in the performance category. The 3GHz and lower versions are stuck on a 667 FSB, and are going to suck wind. Intel looks to be keeping to the standard pricing ramp here, so there are no perks for the people who get a 3GHz. Moral, stick to the higher bin parts or you will pay in performance, but those are going to be quite fast. µ


Share this:

blog comments powered by Disqus
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

INQ Poll

Heartbleed bug discovered in OpenSSL

Have you reacted to Heartbleed?