P4 Clock Throttling Revealed

[Guest]

some clock throttling going on in this Hardware Unlimited story. It appears they saw a major degradation in Quake Arena scores and it is reproducable. There goes *ntel's best benchmark.

http://www.hardware-unlimited.com/reviews/athlon133ddr/index11.shtml

From reading the review, these results are far from normal and way out of context from what we have ever seen before. As a sanity check, we have included a partial review of the AMD Athlon 1.33Ghz, which we plan to benchmark more thoroughly and post a follow up review soon. On the other hand, we uncovered some substantial evidence regarding a possible throttling case with the Pentium 4. Moreover, we did not reach this conclusion in just a few hours of testing. We have been working on this nearly day and night for the past two weeks. When the original wave of questions and comments regarding our findings in our Pentium 4 1.7Ghz review came in we were much more intrigued by our previous unexplainable findings and looked for a method to proceed and further test our results.

This is where John Evans, Bert McComas, and Van Smith came along and helped us. John Evans supplied us with the throttling program and Bert and Van were nice enough to reconfirm our findings and offer a few suggestions in the way of testing. We used production CPUs (not engineering samples) certified power supplies and extremely high quality copper base heat sinks supplied by Intel for use with the 1.7GHz processor. As you might have noticed, I have been much more dormant over the past two weeks on the site as far as news, articles, and overall contribution has been concerned due to this very time consuming issue we are dealing with.

 

[Raystonn]

Interesting article, but it has a flaw. There is no 'software throttling' in the Pentium 4. That would be in the motherboard firmware or in your operating system. The hardware throttling that the Pentium 4 supports cuts the clock frequency exactly in half. Thus if performance is above the throttle-measured specs you can be assured the Pentium 4 is not operating under the thermal protection hardware throttling feature. It sounds to me like they have a buggy motherboard, or the motherboard's BIOS was incorrectly set. It would be the motherboard doing the throttling in 12.5% steps. The CPU only supports full power or 50% throttle.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =

 

[Guest]

Even the author speculates that hardware throttling is <b>not</b> the culprit.

Even after all of this testing, I can’t say for sure that throttling is occurring. It doesn’t look like the automatic hardware throttling is the problem since John Evan’s program should have detected this. It could be that the problem is software driven throttling, and I think that’s what it is.

Sounds like a motherboard issue to me.


(A)bort, (R)etry, (G)et a beer?

 

[Guest]

lets hope its a bad motherboard. Otherwise the 2Ghz P4 0,18 will have many problems.

 

[Raystonn]

The maximum temperature used as the trigger point for thermal protection is set differently for every released speed of the Pentium 4. Thus, the 2.0 GHz CPU will be allowed to reach slightly higher temperatures than the 1.7 GHz CPU without any hardware throttling. As it is, the temperature point for hardware throttling is so high it should not reach that temperature even if a 1.7 is overclocked to 2.0.

As far as motherboards go, the best performer is the Asus. So we can just avoid the problem with that motherboard throttling completely.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =