[R.Bauer]

Hi,


Anyone tried to underclock their processor, and what was the result regarding CPU temperature? Heat is the biggest enemy of a silent CPU-case, so lowering the cpu speed lowers the heat dissipation, and the need for a lot of airflow. Now cpu's are becoming so fast, we do not need the full cpu speed of a 1.4 gHz Athlon anymore to 'just' play a DVD-movie. We now can have the luxury of buying a 1.4gHz Athlon, and run it at 100MHz frontside bus, in stead of 133Mhz front side bus. Let me know the drop in CPU-temperature. (I 'only' have a 900MHz Athlon, so I can not really try it myself) I would like to hear the outcome of this.


Thanks,


Reinier.
http://beamer.xs4all.nl

 

[Hugh Magoo]

I was going to try this, but my Epox board has a vcore of like 1.85. And you can't lower it. So I didn't think the heat would go down if I underclocked. I think the only reason you would underclock is to be able to lower the voltage.


Wait until xcel wakes up. He's done it.

 

[Joe Q]

You may find this arcticle interesting or useful if you are going to try to run your processor without a fan.


It shows that a P4 has built in protection against failure due to overheating but the Athelon doesn't.

http://www6.tomshardware.com/cpu/01q3/010917/index.html


Joe

 

[Ian Wilson]

The fact that you can't reduce the vcore voltage doesn't matter. If you reduce the clock speed the amount of current drawn will be reduced and less heat will be generated.


However you will need to experiment a great deal to find the point at which you can actually use less fans and maintain stability.


Personally I don't care about the temperature as long as its within the manufacturers tolerance and the PC is stable. My HTPC has a stock CPU fan and PSU fan and the case is lined with sound deadening material. It's definately hot inside the case but it's still within the manufacturers tolerance. It runs 24/7, 100% cpu usage as it runs the cancer cracking program - and it never crashes.


Ian

 

[Hugh Magoo]

the vcore on my board is pretty high to begin with so I didn't see much luck in underclocking. I don't think fanless would be an option with a higher clocked Athlon. But it would not be needed either. If you underclocked it a certain percentage, you would be able to keep a very quiet fan on there and not hear it over any of your other fans.


What sort of effects would lowering the FSB have on performance, regarding HTPC uses?


I suppose that reducing the multiplier would be the easiest way to underclock?

 

[Ricardo Cruz]

You can also try the Athlon XP, which requires 20% less power than regular AThlons. I'm currently testing one of these on an ECS K7S5A board, with a Spire CPU cooler. This cooler has a 5000 rpm fan and a copper insert. You can find it here:

http://www.spire.nl/products/5e34b3-H.htm


Well, with NO special case cooling, it's just running the power supply's fan, this system only reaches 45ºC under load, this with a Pioneer 116 DVD and 7200 rpm HD. Also, the Spire is relatively silent.

 

[xcel]

Hi All:


___This will be slightly OT but …


___There are already very quiet but very efficient cooling solutions for the 1.4 GHz T-Bird and the ~ 5 W less output XP 1800+ when using either the Silverado or Alpha 8045 w/ any low rpm 80 mm fan. The Silverado in particular can handle either a T-Bird at 1.2 GHz in its lowest 6 V settings as well as the new XP 1800+ given they both have ~ the same thermal output. These fans are so quiet as to be inaudible inside a std case IMO. The Alpha 8045 can be fitted w/ a variety of low rpm/quiet 80 mm fans and will cool a 1.4 Ghz T-Bird as well but at temps that can approach scary if you ask me. Being below a manufacturers temperature spec when you are measuring the space under the DIE of any T-Bird and XP CPU today is gambling at best. Even the calibration of this method for CPU temps with the AMD CPU’s can run all over the map. I do not know of any boards available just yet for the XP in particular that can actually use the newest XP’s internal Diode for temperature monitoring just yet …


___Now on to the subject … I was playing w/ a PIII 700E clocked to below 500 MHz. w/out a fan whatsoever and it certainly worked. I could run the voltage to just 1.6 V core and everything was stable but anything above 500 MHz, and the temps rose pretty fast. I was using the older but still quite capable GlobalWin FOP-32 HS on the PIII at the time. This HS w/ a Delta screamer placed upon it can and has indeed cooled 1.4 GHz T-Birds but at completely insane noise volumes. The capability to run T-Bird’s is why I chose this HS maybe a year and a half ago for this experiment. Even though the internal temp measurements of that PIII were well below Intel’s spec’s, I did not feel comfortable beating that CPU like that. There are many companies that simply duct the PS fan(s) output to a PIII and have no problems whatsoever with the std. PS fan noise to contend with. With a T-Bird, you can pretty much forget passive cooling as you have about an 8 second window before the CPU burns up w/ the HS placed upon it, let alone being pulled off as the Tom’s HW article did. I am sure 1.33 GHz XP’s clocked to just 1.0 GHz (1.6 V possibly?) via 100/200 MHz FSB may be capable of a passive cooling arrangement given their 41 W or less depending on V Core but AMD CPU’s are simply to fragile for this kind of activity. If you were to use the XP 1500 in this fashion, I have a number of very quiet solutions that will work including the 2 listed above if that is what your after but they work w/ these CPU’s maxxed out w/ the same dB output as long as your case temps/air flows are reasonable. The Alpha 8045 w/ a low dB output Sunon can drive a 1.4 GHz T-Bird to 1.6 GHz without breeching 60 degrees C under DIE temps. When I was using lesser fans, 1400 – 1450 is about all I could achieve before my previously drop dead 64 degree C maximum limit was hit and MBM would auto SD the system.


___Finally, the PIII’s can be cooled passively without much fanfare if you duct from the PS all the way up to their rated frequencies and beyond. The latest .13-micron core based 1.26 GHz PIII Tualatin’s are even better in that they only thermally output what a 933 MHz PIII outputs. The T-Bird’s and just available XP Palomino’s are best air cooled with the best cooler you can purchase and using a quiet fan solution. The Silverado’s in their low flow (quiet setup) will cool the Palomino’s at their max spec (1800 +/1.53 GHz) with under DIE temps below 60 degrees C fully loaded. The Alpha 8045 solution is just a bit better but I had to use faster and faster RPM fans for any kind of OC on a std. T-Bird. The PIIIs can be cooled almost silently w/ an SVC GoldenGate/60 to 80 mm funnel/PC Power & Cooling Silencer, 7 V modded low RPM Sunon sleeve bearing, Panasonic low flow/low RPM Panaflow (I do not have the Panaflow fans in yet to try it so I am guesstimating) … All in all, why risk your brand new $115.00 - $260.00 AMD XP CPU when there are extremely quiet forced fan solutions that are as quiet as you can ask for both protection and at very reasonable costs.


___Good Luck


___Wayne R. Gerdes

___Hunt Club Farms Landscaping Ltd.

___ [email protected]