Fan revs higher after P4 cpu upgrade

[DellDim5000]

I just successfully upgraded my stock 3GHz (530/530J) machine with a 3.8GHz (670). The machine started right up, didn't even halt for the BIOS to show the new processor data.
But now the cpu fan runs at a lot higher rpm level and gets annoyingly loud at high cpu load...
Could that really be just because of the slightly higher operating temps of the 3.8? Or could anything else be wrong with the cpu causing to trigger the fan higher?

[Thandor]

Check out the temps with HWMonitor. If they are high (let's say 63C / 145F +) then either your cooling is inadequate or the fan/heatsink assembly isn't installed properly. In addition to the latter: make sure that you have used just a tiny bit of thermal interface material ('cooling paste') for optimal heat transfer .

[max3]

just an addition - i usually clean old dried out thermo compound clear of the heat sink before applying fresh one.

p.s. this seem to be DEll machine, some dell bios-es have some temp. info in the bios along with non dell mobos, see if anything related to that present, that way no additional software needs to be installed.

p.p.s. everest/astra/sisoft sandra/ etc all show temps and such.

[DellDim5000]

Thanks guys. As to the heat transfer (after having removed any old thermal paste of course) I used arctic silver following exactly the "grain of rice" size application method. But when I checked the paste distribution/contact pattern today, it does not look too good. The heat sink sits on the cpu not much more than maybe half of it's total surface, and unfortunately that affects the important center area (around the core) as well. Whether or not that is the reason for the high rpm remains to be determined, with the old cpu it was not a lot better for sure.

No temp info whatsoever in the bios.

And since this is a P4 lacking a digital thermal sensor, with a proprietary and poorly documented interface of it's embedded controller chip, I have not come across any program able to read out the cpu temp. And I have tested approx. 10 of them so far, including Everest and HWmonitor.

In another forum I read that I can reduce cpu load by adding RAM, is that true?

And again the question, could a defective cpu technically result in high operating temps?

[Qwerty]

[D.8080]

The cpu you have now is known to operate at high temperatures, in fact it was the whistle for no more cpus over that frequency.
It also sucks up a lot more juice (watts).

It does give an extra 30% over your previous prcessore, but at a cost.
Your heatsink is just at his limit in cooling it, and maybe you should buy a more performing one.

Forget overclocking it. It's already an oven on its own.

[Thandor]

[max3]

[Guest]

Sorry for the late reply, perhaps like other guest users I was accidently temporarily blocked from this forum.

In the meantime I have come to the conclusion that there is nothing wrong with the cpu, but the particular heat sink design of this computer, maybe just this very machine of mine, is unable to dissipate the heat sufficiently that the 3.8 generates, which may have even caused the processor to even throttle down during certain operations. Which would perfectly explain the impression under which I ended up when I eventually went back to the 3.0: this one appears to run smoother, maybe faster, at least not noticeably slower (with all kinds of applications) and brought the fan back to run at an almost constant, low, inaudible speed.

Prior to that I took the 3.8 out and put it back in several times, playing with the heat sink trying to get it to a better contact to the cpu, but in vain. The differences were insignificant, and more thermal paste where there was poor contact seemed to do nothing. Amazingly though, for the 3.0 it appears to be "good enough". I would never have believed the difference between both cpus was so extreme in regards of heat regulation.

And no, I am certainly not going to overclock it. I wouldn't even know how to do his right now.

Cpu load reduction by adding ram, this theory wasn't backed up by any explanation whatsoever.

I may look into better coolers next chance, and maybe give it another shot with the 670. Any suggestions for what might fit into this Dell Dimension desktop?

And as far as cpu temperature information, lately I had tried Hwinfo32, Everest, CPUID, Core Temp and Speedfan (a few more some time ago which I can't remember right now). None of them was able to get the cpu temp, which got me to contact the developers of the above ones asking how come. Well, Core Temp excludes "older" cpus such as the P4 in their faq right away. Hwinfo32, CPUID and speedfan admitted to the problem, said they kept working on it but were basically referring to the issue stated before. No feedback from Everest yet.