• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Prescott is HOT

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
This thread disturbs me a bit. I know someone (ok, I helped them pick out the parts) for a P4 system a while ago. MSI Neo2-P Platinum 865PE mobo, Northwood (not Presshot) 2.8C (with HT) 800Mhz FSB CPU. ThermalTake "Spark 3+ Cu" HSF. I grabbed that for $15, instead of the $40 HSF the person was originally looking at, because I didn't think a super-powerful fan was that necessary, and the HSF claimed that it was good for P4 CPUs up to 3.0Ghz, and this was a 2.8C, not planning on being overclocked.

Well, to make a long story short, idle temps of 50-55C, load temps of as high as 70C, yes, system starts crashing and gets unstable. (Whatever happened to thermal throttling? MSI?) Yes, I flashed the newest BIOS for the board. Yes, I cleaned the phase-change pad off, and re-applied thermal grease. Twice.

The case in question is a heavily-loaded AOpen HX-09 full-tower. Built like a tank, but very poor airflow.

For a while, I thought that perhaps he had recieved a Prescott CPU instead of a Northwood, but WCPUID and CPU-Z confirm it's a Northwood. Nominal voltage is 1.525v, but it tends to bounce around a bit under load, not sure if that's normal either.

Ideas? Did he just get a bum chip or something? Seems stable otherwise, at idle.
 
What's ur mobo? Abit? My IS7 with Prescott 3G kept shutting down until i turned off Smart Fan Enable in the mobo control panel. Now my CPU is ranging from 55-65 deg.C with no more problems.

 
Jose's input was good and valid. But with your budget, you really should have gotten an Athlon XP (mobiles are good and cool) or an Athlon64. A good deal for an overpriced processor is not a good deal. Selling the CPU/mobo and getting an Athlon (something) is not a bad idea either.
 
Originally posted by: Mail5398
I have got a 3.2 ghz prescott and it is running at 42 degrees celcius. It gets up to 50 degrees celcius under a load (playing Battlefield Vietnam). I am using the retail fan/heatsink.

Likewise, my 2.8e@3.15 runs the same temp on retail cooling.
 
Great, another "bash the Prescott" thread.

Running my 2.4GHz Prescott at 3GHz at default Vcore, 40C idle, 50C load, using a Zalman AlCu quiet cooler. Using the stock heatsink, it was 10C higher, but still completely within spec. Use a copper heatpipe like a Thermalright SP97, and you'll probably see load temps around 40C.

If you use the stock heatsink, FOR CHRIST SAKES, scrape off the thermal pad and use any thermal paste (AS5, cheapo white paste, etc.) and you'll see an immediate drop in temps. Thermal pads are far and away the worst thermal interface available.
 
Originally posted by: rdubbz420
Originally posted by: Mail5398
I have got a 3.2 ghz prescott and it is running at 42 degrees celcius. It gets up to 50 degrees celcius under a load (playing Battlefield Vietnam). I am using the retail fan/heatsink.

Likewise, my 2.8e@3.15 runs the same temp on retail cooling.
I can't believe people still don't understand that different motherboards make for a complete apples-&-oranges situation. Calibration isn't the same even across all boards from the same maker, or all BIOS revisions for the same board necessarily.

Point in case... nForce "classic" boards from Asus and Abit. Same chipset, CPU, voltage, heatsink, etc will give you temp readings of about 42C on the Asus and closer to 70C on the Abit. The real CPU temp, of course, is the same on both.

Another: MSI recently released a BIOS revision that magically lowered the temp readings on their Socket754 board, IIRC. Did the actual CPU temperature change by about 10C? Or just the reading?

Be responsible when comparing your temps to others' temps.
 
RE:"I can't believe people still don't understand that different motherboards make for a complete apples-&-oranges situation. Calibration isn't the same even across all boards from the same maker, or all BIOS revisions for the same board necessarily."

Really, I though all motherboards were calibrated the same and ALL read the real CPU temperature.
Therefore when someone says their Prescott is running at 42 C and everyone elses is running at 65c all the ones at 65c must be wrong...


Now, I have some hurricane ravaged swampland in Florida for sale if you're interested.


Mac
 
Originally posted by: Macro2
RE:"I can't believe people still don't understand that different motherboards make for a complete apples-&-oranges situation. Calibration isn't the same even across all boards from the same maker, or all BIOS revisions for the same board necessarily."

Really, I though all motherboards were calibrated the same and ALL read the real CPU temperature.
Therefore when someone says their Prescott is running at 42 C and everyone elses is running at 65c all the ones at 65c must be wrong...


Now, I have some hurricane ravaged swampland in Florida for sale if you're interested.


Mac
Muwahaha! :evil:

movingincircles, try www.arcticsilver.com or just use your common sense. 🙂
 
Originally posted by: movingincircles
Hi!

I bought a retail Prescott cpu with boxed fan, didn't need to apply paste. Anyways when I play counter-strike source for 5 minutes, my comp shuts down from overheating

cooling solutions?

better cpu fan?

HOW TO COOL THE FRICKEN PRESCOTT???

Prescotts usually run hot
 
For a while, I thought that perhaps he had recieved a Prescott CPU instead of a Northwood, but WCPUID and CPU-Z confirm it's a Northwood. Nominal voltage is 1.525v, but it tends to bounce around a bit under load, not sure if that's normal either.

It's worth keeping in mind that heat sinks and fans do not lower a CPU temperature TO a particular value, but rather BY a particular value.

Just because someone who is in a climate controlled apartment that is at 21.5C precisely has a particular chip and cooler running at 50C doesn't mean that everyone's going to get that.

First step - what's the ambient temperature ? 8)
 
Originally posted by: movingincircles
noob here, never applied thermal paste

any souls want to point me to a 101 tutorial?

I don't know how far you've got with your problem... Just a few points:

Your stock heatsink probably had some acrylic phasechange heatinterface material on it. That's fine, but only good for one mounting. If you remount the heatsink, remains must be gently removed with the help of isopropyl alcohol. Oh, - and don't pull off the heatsink with lots of force! Twist it a very little, patiently, until it releases.

Temperatures into 70C with Prescotts seem almost "normal". But not desirable. However, I've surrendered to the view that you have to let them go up to ~66C. So set your shutoff higher than 60C. 60C shutoff is a bit unreasonable with a Prescott. This may solve your problem.

What can you do to get temperatures down? Copper! Normally, Aluminum is fine and on a low power CPU copper won't make much of a difference. But copper makes a good deal difference for a really hot CPU.

The nature of cpu cooling is like this:
CPU(temp4)--Paste.transport--Heatsink.Base(temp3)--Heatsink.transport--Heatsink.Fins(temp2)--transfer--Cooling air(temp1).

A certain power output (energy/time) of heat has to be transported through this chain. If we consider the case of sustained CPU load for 40 minutes or more, the temperatures along this chain will have stabilized. So that the energy/time that is transfered to air, by temp2-temp1, is equal to the energy/time that is output by the CPU!

The energy transfer is driven by the difference between temp2 and temp1. So temp2 will increase until the difference is high enough to drive the transfer of the full power output. More power, requires a higher temperature difference to drive the transfer of more energy/time.

The very same then applies to the transport through the heatsink. Temp3 will rise until the temp3-temp2 difference is high enough to transport the energy. It's like you have water tube, and need to transport a certain volume of water through it per minute. You will need to increase the pressure at the input end until you get the sufficient flow. Here's where the copper comes in. Copper is a wider gauge tube. So it will require less pressure to accomplish the required flow.
Aluminum is a very good heat conductor. It's thermal conductivity is 238 W/mK.
But copper is much better. Coppers conductivity is 400 W/mK.
So with copper, we will have a lower temp3-temp2 difference driving the energy transport.

The same thing again then applies to the interphase paste layer, temp4-temp3.

The entire chain is fixed from the outside, temp1! We then desire the least *steep* temperature ladder possible, to arrive at a low CPU temperature, temp4.

The thing you can do about temp1 is sufficient airflow through the case.
The thing you can do about the temp2-temp1 step, is a more vigorous airflow, higher CPU-fan speed.
The thing you can do about step temp3-temp2 is copper and/or clever design, minimizing transport distance.
The thing you can do about step temp4-temp3 is to get the thermal paste layer right, which basically is as thin as possible.

Arctic Silver 5 is a fantastic paste in terms of thermal conductivity. However, it is not possible to get AS5 very, very thin. Which is why, with very smooth and flat surfaces on heatsink and CPU-heatspreader, white silicon paste is preferable. Personally I think it's easier to apply as well, but that might just be habit. You should be able to get silicon paste many times thinner than AS5, and AS5 is not good enough to make up for that. If you need to fill gaps however, AS5 is outstanding.

I have made a little spatula, cut from an old plastic membership card. It takes very little silicon paste, maybe 1/4 - 1/3 of a ricegrain. But I usually get too much, (ketchup) so I don't have a real check. I then carefully spread it on the CPU, very, very thin and as evenly as I can get it, with the spatula. It's somewhat delicate work. It should be thinner than you think it should be 🙂

Arctic Silver have instructions on their website, for applying AS5. I think you basically leave a glob on the cpu and squeeze it out with the heatsink, but you should check.

Copper has an additional advantage. It has fantastic thermal capacity, so it works like a 'reservoir' of cooling. With shorter durations of power output, temp3 will rise slowly and maybe never reach the sustained level. If the cpu temperature is a curve with tops and valleys, copper will smooth that curve, cutting off the tops.


Pitfalls:

Thermal pastes are usually quite toxic, straight through the skin! So don't touch the stuff! (If it weren't for the very special purposes, the stuff probably wouldn't be legal.)

Copper heatsinks are actually *too heavy*. You should avoid transporting your PC, and take great care not to subject it to any bumbs.
Zalman 7000 AlCu is a good compromise. It has copper through the center, where it's most critical, and achieves lower weight through use of Aluminum for the rest.

(Also, in my experience, good heatsink design, are those 'conventional' looking ones, that have a wide bottom copperbase that distributes and transports the heat to the bases of the aluminum fins.)
 
My room temperature is normally around 21 degrees celcius. (70 degrees farenheit.) I have got a 120mm intake fan in the front. An 80mm fan located directly over the cpu on the side and two 80 mm exhaust fans located on the rear. I measured the temp inside the case and compared to what the motherboard was reporting and it was 1.5 degrees celcius different. I do not have a way to measure CPU temperature. I can measure the temperature of the heatsink, but how do you measure the CPU temp without using the number the motherboard gives you? The heatsink temp was less than the number the motherboard reported for the CPU.




 
Back
Top