Can the mechanism be manually disabled, or at least adjusted? If that was done, would the Pentium 4 1.7 GHz have crashed running Quake 3? Why is your implication a "strong" one?
Can't say for sure. You would think there would be some kind of bios control. How about when you get your P4 wok station, you disable this feature and run Quake 3 and come back and tell us if you smoked the system or not? Agreed? Certainly more understanding is needed here. I must say that I'm very surprised that this throttling mechanism has been found on commonly used software even in what is not supposed to be the fastest system (assuming they release a .18um 2Ghz part) If this becomes a large issue and can be traced back to the P4 design (specifically the double pumped ALU) it casts a dark cloud on P4's future. Intel needs to be totally up front on this issue. I doubt it will go away on its own and if it has much substance, AMD will be all over it in spades (especially if equally performing Palominos don't show the effect)
There were only two confirmed cases of throttling in the article, and one case which even Bert and Van admit cannot be root-caused to throttling. And with all of the time Bert and Van spent on this issue (mostly Van, I believe), you'd think they can successfully duplicate these cases using a wide variety of processors out there. But so far, they've only found a couple of isolated cases.
Simple question for you. When is clock throttling on common software an issue with you? On 1/10 systems? 1 out of 100?
1 out of 1000? less?
I'm sure they're investigating this as we speak. Like I said before, the release of the 2.0 GHz Pentium 4 is on the line here, and surely Intel does NOT need another embarrassing situation on their hands.
Let's wait for more study on the subject before jumping to conclusions here.
I agree and that has been my point all along. Much more investigation is needed. But again, I'm disappointed that this stuff comes up only thru a couple of geeks investigation. Does Intel depend on outside sources to characterize their processors? Well, they better not. In this case, they clearly must know the implications of clock throttling since it was designed in to the processor. But, by the way this thing is unfolding, it makes them appear that they are trying to hide it. I hope we see an up front detailed explanation from Intel in the near future.
THE WATSONYOUTH |
