P4: The center of your digital hotplate


— 1:25 AM on May 17, 2001

A month ago at InQuest Marketing Research, its principal contributor, Bert McComas, made a rather sweeping statement claiming that the P4 suffered at the hands of clock throttling. This claim was a tad misleading, and evoked a fair amount of disbelief and rebuttal from the community as a whole. Some even went as far as to brand InQuest as horribly biased.

Well, when one's integrity is questioned, there is only one thing to do: analytically and thoroughly tackle the subject and prove your point. The follow up, "The War Escalates: Athlon4 takes on Pentium4", doesn't do much discussing of the Athlon 4, but it is filled to the brim with Pentium 4 throttling information.

The conclusion? Throttling is a very real function of the P4 processor. Just as it is a function of the P3, and according to this article, the Athlon 4. If that throttling takes place, the P4's performance dives rather ungracefully. There are some caveats to all the information provided, though. These caveats weigh in on the good and bad side of the equation, but if you look at the data that's presented out of context, one would think that this throttling will happen all the time.

Thermal diodes are not accurate. You can sometimes find discussions about the wildly varying temperatures reported by the diode in Intel's forums. Even Artic Silver has an in-depth article on how to determine the margain of error in your diode in order to compensate appropriately. My point? If you get the bad luck of the draw and your P4 diode is too many degrees off in the forward, you could experience premature throttling. Furthermore, as evidenced in some of Mr. McComas' testimonials, slight errors in the application of cooling methods for the P4 could cause the same.

Personally, I still think this possibility is remote. These worries can be nipped in the bud forever by BIOS updates from mobo manufacturers that raise the diode's throttling function to a higher temperature point, or by giving the user the ability to choose a lower temperature point or to disable throttling altogether.

Really, though, with heatsink fan solutions such as Agilent's ArctiCooler product now available for the P4 Willamette and Foster processors, I'm not quite sure what the fuss is about. I believe the cooling solutions available are adequate. Yes, even to handle the hotspot issue—as if the Athlon's triple pipelined FPU doesn't turn into a hotspot during Q3A. Even by admission of one of the testers in the article, correcting a mis-aligned fan in his Dell system, which likely uses cheap ducting, made his throttling problems go away.

The tests provided by InQuest were created alongside a tool that forced throttling. While these results are important to our understanding of a potential pitfall of the P4, we should be careful to avoid making a mountain out of a mole hill. In the mean time, you can debate whether this added complexity is a value (CPU/investment protection), or a pain (additional configuration and attention to detail).

Thanks to Sam Lindhorst for the link.

 
   
Register
Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.