I recently bought this for socket AM2, 65W. I suppose I was a fool to think the stock HSF could be trusted. It's a small, aluminium piece of junk with a 2900 RPM fan. I don't even think the fins are skived. Anyway...

I used the default thermal pad/tape that was already on the HSF, and to my alarm, I saw the temp approach 80 C while under stress. And this is at normal speed. I only let it run like this briefly once I discovered it (probably 10 mins gross, at max). But even while idle on my desktop, it could reach 70 or more.

I quickly decided something needed to be done about this. So I ordered an Arctic Cooling Freezer 64 Pro, and Arctic Silver 5 grease. Now starting temp is around 50 C, with sustained idle temp between 55 and 65 C Under full load, I've seen it climb to 71 C

I looked up AMD's PDF on these, and they list a max temp between 68 and 70 generally. If I let it run at 71ish long term, am I at risk for damage? Also, is there any way to determine if there has been physical damage, short of a complete burnout, or frequent crashing/lockups? Example : could previously absent frame rate loss in games be an indication?

I guess the only other thing... could the temp monitor on my motherboard be providing inaccurate readings? I have a Biostar based on Via K8M800 chipset, with a 20 pin power connector. I'm getting my readings from ITE Smart Guardian software (and the BIOS reading are usually similar).

I'm not too fond of the Arctic for a couple of other reasons... the sink doesn't touch the entire core (it leaves about 1 or so milimeter of space on at least one side). It's very big and heavy, and the fan RPM is only 2300 max. I'm considering getting either a ZeroTherm BTF90, or a Zalman CNPS7700. Any advice on either of these 2? I'm pretty sure the ZeroTherm will fit in my case.

Someone asked me how my case ventilation was, so I'll post this info with something else that may be pertinent too. Remaining system specs:

EnerMax Noisetaker EG475P-VE (additional 12 volt, 4 pin connector plugged in)
4000 RPM system exhaust blower in PCI back panel
Vantec 80mm case fan under PSU

Honestly, most of the air being blown out of the case isn't that hot - in fact it's barely warm.

What's going on here? Do other people's 4800's get this hot?

I have a fan similar to the 120mm all copper Zalman "mushroom" (CNPS7700) on my FX-60. It keeps it VERY cool. I can't remember the temps right now, but I think that they were under 50C at max load.

Still, that's surprising that your 4800 is so hot, considering it is a 65W part and mine is a 130W! Do you have good airflow through the case? I've never used the stock cooler (the one for my old X1 3800 is still in the box), but I was under the impression they were at least decent.

I doubt that those temps would kill your processor, but I would not feel comfortable with them and would probably get a cheap aftermarket cooler, such as this one (the one I have, it is a rip off the Zalman):

http://www.tigerdirect.com/applica [...] u=ULT33050

This one will work on 939's for sure. Is yours a 939 or AM2?

It's an AM2 Case ventilation is fine. I think I'm going to go with the ZeroTherm I just hope I haven't done any damage already. I probably won't get the new heatsink until next week. And I can't stay off my PC for too long... I run my business from it

Well, I had an old A64 3500+, and I think it was the sensor on an old ECS board. Reported idle and max temps around yours (I even went to water!). Tossed out the water 'cause it was a hassle, and then the motherboard died. Replaced it, and I was getting idle of 30c, max at about 45c. Dunno if that might be your problem, though.

Man, unsure why you'd have that problem, I've got mine with a stock heatsink/fan, I'm running a 5200 though, and big side fan, mine idles under 30 degrees.

I think you found your own problem when you stated the heat sink didn't mount flush on the cpu. Something is wrong with your HSF mounting.

I have the same motherboard and my 4200+ X2 is also around 65 degrees and many other users claimed that the CPU is supposed to be cool. I think that it's the motherboard temperature and not the CPU's according to hardware monitor. Use core temp to check your temp. Mine show that the cores are around 15 idle and 30 under load.
P.S. what kind of AGP card do you use?

Yeah, guess that makes me an older timer, huh? Using AGP. I made a significant investment in a BFG, GeForce 7800 GS, 256MB. Over $300. It is OCed stock.

See the thing is, I'm kind of reluctant to buy new technology I'm not terribly familiar with. I haven't been on the main Tom's site in a long time - so I'm behind. Some time back, I kind of saw PCI Express as unproven technology. In truth, it isn't a slot that is dedicated to graphics cards, like AGP. As I understand, it can accept sound cards, NICs, etc...

See, that's the thing. I didn't want to get a Biostar, but that was the only thing NewEgg still had that was compitable with my current system. They had an MSI based on Via K8M800, but it had a 24 pin power connector, whereas my EnerMax PSU is 20 pin. I really wanted to get an SiS chipset, but it seems AMD is making good quality motherboards now.

I've learned that in recent years, processors are becoming more essential to PC gaming that video cards. Where with most games, you can tone down in game options to compensate for older cards, there's not much you can do about CPU processing (AI, path finding threads, etc...). Upgrading video cards now seems only necessary if the card needs a next generation feature via hardware routes (ie DirectX 10 acceleration, pixel shading 4.0, etc...). Of course, software routes really slow things down, but they still allow the game to play.

As for me not mounting the HSF correctly - no that's not true. It is flat against the core... just on 2 of the sides, it doesn't reach the tip of the edge. The clips are in exact place, and I know before I installed it, that it didn't cover the entire core. That was also only on this Arctic Cooling, not the stock HSF.

Maybe I can just hope the motherboard reading are inaccurate - or perhaps just elevated?

Hey guys, one thing I've noticed is that the Vcore voltage appears to be a little low. I have it set on auto in the BIOS. The readings are hovering around 1.25V While the listed voltage is supposed to be 1.325 minimum, and possibly as high as 1.4 Could this be causing the problem? I did wonder if a 475 watt PSU would be enough.

That is really high. Does it ever crash or anything? What happens if you run Prime95?

Honestly, my guess is that the temp sensor is mest up. If it isn't, though, and you are really getting that high, I would return/exchange it for a new CPU and MObo.

Hasn't crashed so far. But... why would the temp sensor be displaying artificially high readings? Just bad design by Biostar or Via? Or perhaps just a flaw with my particular board? One thing I did find peculiar... there was no black plastic guard over the CPU socket on the board (and there clearly is in the manual). I bought it new, retail from NewEgg - so I know it hasn't had a previous owner. The I/O plate and other accessories were unopened, still inside plastic bags that can't be resealed.

Maybe the sensor just got dinged around a bit during shipping or something? If so, that appears to be the only thing malfunctioning - as the rest of the board is fine (though my hard drive is running a little slowly)

If this is the case though, I have no way of knowing what the actual temp is. Any way to get another sensor in there, or something equivalent?

Wow, I've owned the x2-4200 and the x2 5600, the 4200 run about 28c, and the 5600 ran about 34 celcius of stock fans, when i replaced the stock fan with the arctic cooling 64fan, my temps wnt to about 28c on the 5600, so something your doing is wrong, I'd clean off everything and start over, the temps are way hot.

That's a bit high, I have an A64 X2 5600 running at 3015Mhz (215x14).

I get 38 degrees on stock HSF, and at max, mine's at 56 Celsius, by running Folding@Home on SMP clients.

Well, in my experience, I get that temperatures more naturally at Pentium 4's, and when they get to that temperatures, that are so hot to the touch I am physically unable to touch the heatsink for more than 3 seconds, while I can touch the 56 degrees A64 temperature.....

Gene over at AMD says it is quite possible I could be getting inaccurate readings, especially if I have a Brisbane core (ditto). He said either that or the MB's diode sensor could be defective - or not translating temps properly. But if I really want to be sure, I should put my finger against the heatsink and see how hot it is. If it burns me, the temps might accurate - if it is only lukewarm, then it is definitely a prob with the readings.

Stangely enough, he discouraged the use of silver thermal compounds because of their "high electrical conductivity" He suggested ceramic or silicone instead.

Well, the silver compounds are fine as long as you are careful and don't get any on the MB or processor pins.

I'd do the touch test like he suggests. The more I read your posts, the more I think that it just an inaccurate reading. Those temps are insane if true.

Here's another test for ya. Turn the computer off fo an hour or two, then turn it on and go straight into BIOS. If it reads 70-80C in there already, then you know it is just a misreading.

I've never seen it below 50 C in the BIOS. Even after my tower has been turned off (mind you, turned OFF, not sleep mode) over night for several hours. Same thing for the ITE software.

Maybe some not so bright manufacturers accidently label F degrees and C degrees?

First thing, see if taking the side off affects temps. If it does, you need to balance air in + air out.
Are you sure you tightened down the hsf all the way? Can you rotate it at all?
The old finger on the hs is a good test. I borrow an IR temp sensor, but a finger is okay.
If you cant rotate the hsf and the hs is not too hot to the touch, you probably just have a poor sensor calibration.
It's important to remember that a chip will get flakey before it gets damaged. I use F@H to check stability. If it can run @ 100% for hours on end, your temps are fine.

Where can I find an IR temp sensor, and what do you mean by flakely? Also, what's F@H?

Maybe one other thing that contributes to the current theory of inaccurate readings... For those 10 mins I did run it using the stock HSF, and it was reading damn near 80 - even if that didn't result in permanent damage, would it not have at least crashed the proggies I was running to crash? Because they didn't.

My Vantec case fan is directly behind the heatsink. The air is it blowing out is in the same direction. I'd say the air is a little warm at worst, but that's it. Shoot, the air my PSU is blowing out is warmer than that from the case.

Hehe, that would be funny if that was F instead of C (its a little low to be that I think, though). It wouldn't be the first time units screwed stuff up.

I just grab an IR thermal unit from work. One that I use sometimes, costs ~$5000, but some of the others are <$100. Know anyone at a large electrical maintenance company? Some tool rental places might carry them as well.
By flakey, I mean you get errors.
F@H is folding @ home, it's a distributed computing program. It would take too long to explain more, just google (yahoo) it. Any error at all will cause the program to restart, so it's a great stability program.

That may be a good sign, but a lot of today's progs are fairly forgiving. I use F@H, but Sysoft Sandra Burn-in works, as does Prime 95, and a bunch of others, because they respond quicklyto instability, and put a lot of stress on the chip.