Most of you already know the talk of the town: The two different vidoes on how AthlonXP CPUs cope with too much heat.

The first video was done by Tom's Hardware, read the original article and the recent follow-up. It shows how a Athlon and a AthlonXP get destroyed by overheating as soon as the heatsink is removed.

The second Video is AMD internal and was released (amongst other sites) by AMDZone and commented at sites like VansHardware, 64bits.org or in the AMDZone Forum. It shows a Palomino surviving taking off the heatsink.

So what's the problem?

A LOT of people are now saying Tom's Hardware wasn't doing the job right, while AMD was. A lot of people do dislike Tom's Hardware and say so while attacking this article and at the same time only having marginal knowledge on the topic. Furthermore, a lot of people are telling bogus info and non-verified rumors.

This article is an attempt to set the facts surrounding the problem, the article and the videos right. I will tell the source for every piece of info, as well as mark my own opinion as such.

First of all: my comment on the the argument

If AMD shows it works and Tom's Hardware shows it doesn't work, was Tom's Hardware doing something wrong showed a fake, was the mainboard doing something wrong? No. Both Videos are not contradicting each other!

I will back up what I say now with facts and sources later. The problem is, you can't compare the two videos. You have to know the background of these two videos.

The first Video(by Tom's Hardware Guide, furthermore referenced as THG) was done to show how different CPUs and Mainboards cope with the situation that the heatsink is taken off. The Palomino was used on a mainboard that definitely does support the thermal diode and that does have onboard circuits designed to throttle the CPU in case it's overheating. It fails to save the CPU in this extreme case of no cooling at all, though.

The second Video by AMD shows another Mainboard, wich has been fitted with a temperature controlled shutdown circuit. Basically they took a commercially available mainboard, built a device that reads the CPU temperature through the on-die diode and cuts the power if this value is over 85°C. Fitted with this EXTRA hardware, the Palomino survives.

So Tom's Hardware Guide was NOT faking, and they were NOT doing anything the wrong way. The AMD Video does NOT contradict the THG Video.

The only two things to say against the THG Article are:

- The "issue" is not as big an issue as THG says it is (being able to survive without heatsink is not one of the core qualifications of a CPU)

- THG was wrong in telling that the diode itself was the cause by being too slow to react. It is wrong that it can only detect a change of about 1°C per second.

- The Pentium 4 related things I will not discuss here, but the low temperature measured surely is not right.

So now for the list of facts that are gotten wrong all the time when discussing the Videos:

Now to get some things clear that are told WRONG in articles and discussions on this topic:

Primary FACT : The Videos can't be compared because AMD uses extra attached hardware to cut the power in case of overheating, while THG relies on the onboard thermal management. THG Did not disable thermal shutdown in BIOS or something like this (it's not possible on D1289 to disable thermal management)
Sources :
- Info from FSC in THG article "Siemens assured us that the thermal protection circuitry is definitely working on their motherboard"
- AMD Video tells us MAX6512 is used, that is the extra circuit added to the GigByte Board used

1. FACT The Mainboard used by THG (Fujitsu Siemens Computers D1289) DOES have full support for the thermal Diode of the Palomino Core. In fact that's the only way it can tell the temperature, it has no other sensor as it was developed for Palomino CPUs.
Sources :
- I own this mainboard, it DOES use the thermal diode, I used it in this cooler test.
- FSC Website on the D1289 tells about thermal management
- Info from FSC in THG article "Siemens assured us that the thermal protection circuitry is definitely working on their motherboard"

2. FACT The D1289 DOES incorporates thermal management designed to throttle the CPU in case of overheating. It has a dedicated microcontroller on board regulationg fan speeds and throtteling the CPU if it gets too hot. This is independent of the CPU, BIOS or OS!
Sources :
- I own this mainboard, it DOES have thermal management, I saw this in my cooler test plus you can control this via the FSC Mainboard software SystemGuard
- FSC Website on the D1289 tells about the thermal management
- Info from FSC in THG article "Siemens assured us that the thermal protection circuitry is definitely working on their motherboard"

3. COMMENT Neither the thermal Diode nor the onboard controller are too slow.
The real problem is that throtteling an AMD CPU does not keep it cool enough to survive. The only way to keep it alive is to cut the power, which would have to be done by special hardware - just as the one shown in the AMD Video.
Sources :
- I own this mainboard, it DOES show jumps in temperature very fast. You can visualize this via the FSC Mainboard software SystemGuard. Tests done by me (which will be on this page in a few days) show that the throtting DOES work (causing performance to go to about 1/3 to 1/2 of normal) but temperatures are not dropping significantly, as it is the case with Intel CPUs when throtteled
- AMD sees this as an option and is about to recommend using the circuit shown in the video on all mainboards.

4. COMMENT Neither AMD nor Fujitsu Siemens Computers have ever claimed that the AMD cpu can survive without any form of cooling. The protection function of the thermal management is meant for cases when the fan stops working or the heatsink is not 100% correctly attached (no thermal grease, left the protective foil on phase change pad etc.) In these cases it WORKS. See the cooler test article: there the CPU is kept at just below 90°C with a seriously undersized heatsink...

5. COMMENT The thermal diode in the palomino is no "half hearted" attempt on thermal protection, but simply a leftover from the mobile Athlon, where having accurate temperature measurement is a must. As the AthlonXP has the same core it features the same diode, too.

6. FACT The chipset has got NOTHING to do with supporting the thermal diode or not. The diode is accessed through two pins on the AthlonXP which have to be connected to a sensor circuit. So no new Microcode on the CPU can shut it down, as the CPU itself does not know how hot it is. Only the external circuit can tell this.
Sources :
- AMD Data sheet for AthlonXP, Pages 23 and 37

7. COMMENT I am quite sure AMD would not have adressed this "issue" without the article. This "issue" was no issue to AMD or Fujitsu Siemens Computers (or noone else I know of) before the test, as being able to work without any cooling is not a feature that is demanded or specified for current CPUs. If, as it's the case with intel, this works - that's a plus. But if it doesn't then it's just normal. You wouldn't complain if your car malfunctioned and went off the road when it lost one wheel - You wouldn't demand an extra wheel in case one of the four regular ones falls off, would you?
By the way, as soon as the cooler comes loose, you have a lot mor problem than your CPU possibly getting burned. Your mainboard is useless due to the fact the noses holding the cooler broke off, the cooler is dangeling in the case shorting out your graphics card or whatever ... think of this.

Comments on certain articles

http://www.64bits.org/cpuheat/cpuheat2.htm : "Apparently, the board used in the test was a older one which did not know about the thermal diode and hence simply ignored it." This is WRONG. See points 1 and 2 above as well as below.

"the Siemens D1289 does have a onboard Thermal Protection Circuitry that uses the onchip thermal diode on the Palomino. The Asus A7V266 also has such circuitry, but unfortunately it wasn't used in the test" The protection is different. D1289 does it in hardware, A7V266 in Software (BIOS). If the CPU overheats quickly the CPU crashes, the BIOS can not shut down.

http://www.vanshardware.com/index.html " "AMD Burn" Motherboard Predates THG Video
According to inside sources, development of the motherboard used in the so-called "AMD Burn" video predates the infamous "Fallen Off Heatsink" article on THG." The two videos CAN NOT be compared. See first point above.

"In all scenarios, the Palomino comes out unscathed. A similar though much less thorough test came out with unsurprisingly different results at Tom's Hardware. Ouch! Looks like dispensing bad medicine can result in a mouthful of looser teeth. Good job Ben & Joe, perhaps you can also give THG a crash course in analyzing computer technology" see above

Conclusion

All in all I would be glad if the new AMD solution is adopted by the next series of mainboards. Although it is not a minus point in my opinion if a mainboard doesn't have this shutdown, it's a nice plus if it has...