Please register or login. There are 0 registered and 1213 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s September 23 - 01:39am EDT 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  Clock throttling, a hot issue? 
  Jun 01, 2001, 09:00am EDT 


By: Sander Sassen

One thing that's obvious from our testing is that the thermal management does work and clock throttling is indeed something that could happen to your Pentium 4 system, but only if your OEM heatsink fan stops working and you neglected to apply thermal paste to your heatsink. It is then quickly followed by the thermal protection circuitry shutting the CPU down before real damage is done. In all other cases we were unable to see the clock throttling effect show up in our benchmarks, the CPU would simply not finish the benchmark and in most cases lock up halfway through if any thermal issues occurred. In comparison to our cryogenic cooled Pentium 4, its SYSmark 2000 score is only 0.005% (191/190 x 100%) higher than that of the others. A difference that can be attributed to the increased efficiency of the CMOS process used in the CPUís manufacturing at low temperatures.

As can be seen by the table we used about every combination possible to see if we could show the effects of the clock throttling. The thermal monitor might have been turned on at times and clock throttling might have actually happened, but it did not influence the benchmarks one bit. An aluminum heatsink without thermal paste but a working fan turned in the exact same score as a copper base heatsink with thermal paste and fan. If clock throttling was used to keep CPU-die temperature down to functional limits then it only lasted for a few mili-seconds at most, which apparently isnít enough to influence the benchmark.

Therefore we think that the clock throttling feature is one that successfully prevents the CPU-die from high thermal ramp rates that could damage it beyond repair. Itís a pre-caution and a safeguard when thereís something seriously amiss with the cooling solution mounted to the Pentium 4 CPU. With a proper heatsink mounted, that is able to handle the heat load from the CPU, thereís no reason for concern that the clock throttling feature might affect the performance of the Pentium 4.

Overall, we feel weíve looked at the whole clock throttling issue in a real world situation, under normal conditions, and we can vouch for the Pentium 4 retaining every last bit of performance it is said to offer. However we can see that you might run into problems with a case temperature thatís over 50C, a low-cost OEM heatsink and a CPU fan that is clogged by dust. But even then we doubt that youíll see substantial decreases in performance due to the clock throttling feature, as it simply doesnít turn on for longer than needed, which is probably mili-seconds instead of seconds. But in that case youíd need to seriously think about the heatsink mounted and replace it with one that is able to handle the heat load and/or start looking at other options to bring the case temperature down.

We can only compliment Intel for the thermal management features it has implemented on-die as quite frankly running the same tests on for example a 1.33GHz Athlon CPU would certainly have resulted in the Athlonís core being damaged as it lacks those features and cannot safeguard the CPU from overheating. So in that case proper cooling and monitoring of CPU and case temperatures is Crucial if you want your system to be up and running stable at all times. But to be honest, proper cooling is always something that deserves attention, with any CPU, as frankly spending $400 on a CPU and only $5 on a heatsink doesnít make sense. Good heatsinks can be had for as little as $30 and are worth every penny, especially with high-end CPUs such as Intelís Pentium 4 and AMDís Athlon.

Sander Sassen

1. Introduction
2. Heatsinks and die-temperatures
3. Hotspots, and heatsink materials
4. Thermal paste, other factors
5. Thermal management, clock throttling
6. Thermal monitor, CPU safeguard
7. Thermal testing, go or no-go?
8. Test results, fail or success?
9. Conclusion

Discuss This Article (9 Comments) - If you have any questions, comments or suggestions about the article and/or its contents please leave your comments here and we'll do our best to address any concerns.



  Related Articles 

A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.