Athlon XP CPU temps... 50C is NOT high!

[Jeff7181]

I keep reading this over and over and over and over... people saying 50 degrees C is too high. This simply isn't true. I'm on my 3rd XP2500, now a mobile, but the other two were regular desktop XP2500's... and ALL of them have run 50 degrees C or higher for months at a time without problem. I've even run one of them at 60 degrees C for nearly a month with no instability or other heat related problems. I also have an XP1700 Palomino that's been running 62-65 degrees C for 16+ months, and I had an XP1700 Thoroughbred that ran 55 degrees C for over 6 months with no problems.

Some people, especially the anti-overclocking group, are under the impression that heat kills processors. That's partially true, but 50 or 60 or even 70 degrees C is not enough to kill a processor. AMD's Athlon XP's are rated for 85 degrees... the Mobile Athlon XP's (Athlon XP-M) are rated for 100 degrees C. As we all have heard recently, the mobile chips are made no differently than regular desktop chips, they're simply selected to be mobile CPU's because they are superior chips that can be run at lower voltages at the same speed, or higher speeds at the same voltages.

I know there is a certain margin of error that must be taken into consideration since AMD's specifications refer to the actual core temperature, not always the temperature being displayed by monitoring software that may be measured somewhere on or around the CPU socket, not the actual core itself. And then there's also a margin of error that should be maintained purely because you don't want to take a chance of overheating your CPU and damaging it. If we assume a 10% "margin of error" factor for the socket temp being read rather than the actual core temp, and another 10% for the "peace of mind" factor, you come up with about 70 degrees C being the high point.

So 70 sounds high to you? Does to me too... but AMD says it's ok. Granted this is at stock speeds, but it's my opinion that if your CPU becomes unstable when it gets warm, it's more than likely due to your overclock being unstable to begin with, and heat only exacerbates the instability. You can argue that the ability of the processor to reach higher speeds decreases as the temperature increases, but the reality of it is you're talking about only a few Mhz difference under these conditions, maybe tens of Mhz at the most... which only makes a difference to the most anal of us all who define their status as a human being by their score in PCMark, 3DMark, or some other synthetic benchmark.

Anyway... to end my rant... people, stop recommending that someone upgrade their cooling setup if their processor hits 50 degrees C... it is NOT necessary. If you're having problems with your CPU running at 50 degrees C, it's more than likely not due to the temperature, but rather some other weak link in your setup.

 

[CarpeDiem99]

Good read

 

[Elcs]

I agree that 50C isnt high.... Id even go as far to say that 65C is as high as Id let my processor run at constantly.

But... if people intend to overclock a processor to its max, assuming safe voltages... and they are idling/loaded at 50C at stock with current cooling, then it may be benefitial to upgrade cooling to either achieve a higher overclock more easily or be safer when doing so.

I do agree that some advice given out in the forum in particular is completely biased toward those of us who grab an XP (as an example), throw 1.75V (or higher :frown at it and see how high it will go before being unstable. Those of us who do that probably need high end cooling.

My Barton 2500+ ~ 2.3ghz @ 1.75V actual never goes over 52C with an SLK900A with 92mm Panaflo L1A.... usually 42-44C idle. Comfortable temperatures.... at a nice volume (for me).

 

[Jeff7181]

Good points Elcs. Many people are using Thermalright copper heatsinks from the old SK-6 up to the new SLK947 or whatever the hell it's called... all of which get the job done. The newer ones may bring the temp down a few more degrees, but a few degrees isn't going to make or break your overclock.

I should also mention, for those of you who don't have your case open every month or so swapping out parts, it's a good idea to check your temps every few weeks. Dust will build up between the fins on the heatsink and reduce airflow through it... your temps may rise as much as 10 degrees because of that, even more if it's seriously clogged up.

 

[Jeff7181]

I actually came up with a pretty simple equation based on Tom's Hardware Guide's P4 @ 5.25 Ghz at -196 degrees C, and then I used Intel's specification of 75 degrees for their 3.4 Ghz processor

y = -7x + 3916
where...
x = temp in C
y = speed in MHz

(this is of course assuming that the relationship is linear)

So... based on this, a temperature of 45 degrees C would allow a speed of 3601 Mhz, which sounds about right since that's almost right at the upper limit of the Northwood on default voltage.

*EDIT* Summary... a 1 degree C drop in temperature will allow for a 7 MHz increase in clock speed, lol.

 

[Jeff7181]

Actually, I just found fault in my equation, lol... Tom's test isn't done at default voltage.

 

[funnyfenix]

no clue how the came up with that equation but it sounds interesting

 

[myocardia]

Originally posted by: Elcs
I agree that 50C isnt high.... Id even go as far to say that 65C is as high as Id let my processor run at constantly.

But... if people intend to overclock a processor to its max, assuming safe voltages... and they are idling/loaded at 50C at stock with current cooling, then it may be benefitial to upgrade cooling to either achieve a higher overclock more easily or be safer when doing so.

I do agree that some advice given out in the forum in particular is completely biased toward those of us who grab an XP (as an example), throw 1.75V (or higher :frown at it and see how high it will go before being unstable. Those of us who do that probably need high end cooling.

My Barton 2500+ ~ 2.3ghz @ 1.75V actual never goes over 52C with an SLK900A with 92mm Panaflo L1A.... usually 42-44C idle. Comfortable temperatures.... at a nice volume (for me).

I personally have owned two XP's that would reboot at or around 54-55C (obviously not die temp, Jeff, but socket temp, like nearly everyone else in the world but you uses ) and I know more than one person who's random reboots with an XP went away entirely the moment they added more cooling. And, I still think that I was right-on with the guy the other night who said he was running 50C in the bios screen!, wanting to know how far he was going to be able to overclock. My computer, which has enough airflow to float a hovercraft rises 2 degrees C, by booting into Windows, and doing absolutely nothing, which means that his idle temps are ~52C! Or 62C, for you, Jeff. Would you start trying to overclock your computer, if it idled at 62C (the way you have it setup to read)? And if you think lowering temps don't have anything to do with overclocking, you need to do more research on the subject. Why do you think that having phase change cooling allows a higher overclock on the same voltage?? Anyway, I'm through with my rant now also.

 

[Jeff7181]

You obviously didn't read what I said... read it again, then you can edit the last part of your post

*EDIT* Oh, and I still say if you or whoever you were talking about, experience instability at 50 degrees C, and better cooling "fixed" the instability, your instability was caused by something else that was exacerbated by extra heat. Of course, one possibility could be a defective CPU... since AMD's CPU's are warranted up to 85 degrees C

 

[AWhackWhiteBoy]

your forgetting about the whole point, the lower the temp the higher one can push the bartons,which is why everone talks about temp. 5C change in temp when i got my new heatsink yeilded an extra 150mhz. temp means everything when overclocking the barton.

 

[Jeff7181]

Originally posted by: AWhackWhiteBoy
your forgetting about the whole point, the lower the temp the higher one can push the bartons,which is why everone talks about temp. 5C change in temp when i got my new heatsink yeilded an extra 150mhz. temp means everything when overclocking the barton.

I very seriously doubt it.

 

[AWhackWhiteBoy]

believe whatever you want,i saw the difference it made first hand, if you look at the heat output of the bartons on a chart you'll see at 2.2ghz it curves up exponetially and heat become a major issue. getting my cpu below 55C yeilded much better overclocking results. the cooler the better.

 

[Elcs]

One thing that Myocardia just reminded me of...

When i first got my SLK900A, I said "dont for the life of you put it on your processor backwards". I was very careful... and put it on backwards. My NF7-S Rev 2.0 reported temps of ~43-47C yet my PC would reboot when trying to do anything even slightly stressful, cutting out within a few mins at stock (1700+ T'Bred B).

Once seated properly, my T'Bred B overclocked back to what I had it set to on my Volcano 7+ albeit with lower temps.

While I still stand by what Jeff7181 mentions about people being recommended to improve cooling when it isnt always nessercary, I do believe that the Overclocking/Enthusiast group of users should have access to reading the actual in-built thermal sensor. It has to be far more accurate than a diode on the back of the cpu. I think my case above may prove the point here.... board diode said my chip was fine, yet when correctly installing cooling... it worked fine.

 

[myocardia]

Originally posted by: Jeff7181
You obviously didn't read what I said... read it again, then you can edit the last part of your post

*EDIT* Oh, and I still say if you or whoever you were talking about, experience instability at 50 degrees C, and better cooling "fixed" the instability, your instability was caused by something else that was exacerbated by extra heat. Of course, one possibility could be a defective CPU... since AMD's CPU's are warranted up to 85 degrees C

Three of those were not overclocked even one 1 mhz. One of them was a Palomino, and overclocked 200mhz after lowering the temps (not shabby for a psu-limited Pal.), with not one random reboot again, and the other two are still in use today, also not rebooting for no reason, once the temps were lowered. The Pal. was sold, then sold again, so I have no idea of it's condition today, but I seriously doubt that either of the other two were damaged in any way, since one has been in operation for ~15-16 months and the other over two years, since lowering the temps on them. Take from it what you will. None of those are Bartons, though. I have never owned a Barton (it better get here today), but from what I've seen, the Bartons are quite a bit more stable at higher temps. You realize that just because a company says that 85C won't destroy a processor does NOT mean it will be stable at that temp, just that it will be usable after you find out and correct why it was that hot. And I'm not sure about which part of my last post you are talking about, but would definitely explain anything, or retract what I had said, if it were wrong. Don't get the impression that I'm mad, Jeff. You don't know anyone that's harder to make mad than me, I promise.

 

[Jeff7181]

Elcs, maybe you misunderstood... by "diode temps" I mean the temp read from the internal diode in the CPU core... sounds like you're saying diode and referring to the "socket temp."

believe whatever you want,i saw the difference it made first hand, if you look at the heat output of the bartons on a chart you'll see at 2.2ghz it curves up exponetially and heat become a major issue. getting my cpu below 55C yeilded much better overclocking results. the cooler the better.

Show me.

 

[myocardia]

Originally posted by: Elcs
One thing that Myocardia just reminded me of...

When i first got my SLK900A, I said "dont for the life of you put it on your processor backwards". I was very careful... and put it on backwards. My NF7-S Rev 2.0 reported temps of ~43-47C yet my PC would reboot when trying to do anything even slightly stressful, cutting out within a few mins at stock (1700+ T'Bred B).

Once seated properly, my T'Bred B overclocked back to what I had it set to on my Volcano 7+ albeit with lower temps.

While I still stand by what Jeff7181 mentions about people being recommended to improve cooling when it isnt always nessercary, I do believe that the Overclocking/Enthusiast group of users should have access to reading the actual in-built thermal sensor. It has to be far more accurate than a diode on the back of the cpu. I think my case above may prove the point here.... board diode said my chip was fine, yet when correctly installing cooling... it worked fine.

I agree wholeheartedly, that we should have access to the true die temps, but for the moment, we only have the diode temps, and one of the really easy workarounds (yes, there's a workaround for EVERYTHING with computers ) is keeping your temperatures well below what AMD or Intel says they will run at. I will freely admit that I'm one of those people who is anal about temps, and fairly anal about voltages, more so with giving advice than even with my own stuff. Why? Because I would feel responsible, if someone who freely admits knowing nothing about overclocking were to fry his chip that he very well might not be able to afford replacing, because of my advice. I don't think that is wrong, do you?

edit: Okay, I just thought about something: there is a workaround for everything with computers, except hard-locked multipliers.

 

[Jeff7181]

And if you think lowering temps don't have anything to do with overclocking, you need to do more research on the subject.

That's the part I was referring to... I never said lower temps don't have anything to do with overclocking... don't put words in my mouth I said the small changes in temps people are talking about here ("50 C is bad, 45 C is good") have little or no effect on the stability of the CPU.

I also never said any of your CPU's were overclocked And... call up AMD and tell them your CPU has been running 80 degrees for the past few months, and now it doesn't work, and see if they'll replace it. I guarantee they will if you were using the retail heatsink.

I don't think you're mad either... I think you're reading into what I say WAY too much. I mean exactly what I say, no more and no less =) If I turn out to be wrong, that's fine, but if you're trying to prove me wrong about something I never said, or even implied, that's another story. If you're just trying to make a point to ad to the information here, great, keep them coming, but don't assume I don't know something just because you assumed I was talking about something else.

And one more thing for WhiteBoy... if what you say is true, that reducing your temp 5 degrees C allowed you an extra 150 Mhz, then I guess if I stick my computer outside in the winter where it's 20 degrees cooler than in here, the processor should run about 20 degrees cooler, so I should be able to run mine at 3.0 Ghz at 30 C since right now it's running 2.4 ghz at 50 C... and according to your findings, for every 5 C decrease in heat, you gain 150 Mhz... Hmmmm... taking that into consideration... if I use that same cooling setup Tom had that takes the processor to -196 C, I should be able to run my Barton at... 8.25 Ghz. Obviously it doesn't scale linearly like that... so the question is really, at what point do you start to have to reduce the temp by a WHOLE lot more in order to gain more speed? I say the temp that the CPU is rated for. If you've read anything about how CPU manufacturers "burn-in" CPU's, you'd know that they're run under extreme conditions (very hot, very fast, with lots of voltage... more so than it would EVER see in ANYONE'S PC) to weed out the ones that are defective and would most likely fail real soon... so again... I ask for proof... show me this chart you speak of.

*EDIT* Actually... don't show me... cause it has nothing to do with my original post... you're arguing a point that doesn't have anything to do with what I'm talking about. You're saying you went from 60 to 55 and gained 150 Mhz... I'm saying people who recommend a better cooling solution at 50 degrees aren't correct. Bartons, Thoroughbreds, and Palomino's are all capable of running at 50 degrees C.

 

[myocardia]

Jeff, I'm not going to spend 2 hours (I've got dialup, no choice) searching the internet for something to prove to you that a lower processor temperature yields a higher overclock, but why in the world do you think that spending ~$1000 on a Prometia yields higher overclocks every time than the same chip will run on air cooling? If you honestly think that I'm wrong about that, feel free to waste your time doing the research, but don't be surprised if you find out that I'm right about lower core temp equaling higher overclock. If you don't believe me, and don't want to do the research, unplug all of your fans, without lowering your overclock, then hold your hairdryer close to your cpu on high heat, while running Prime95 on a Prime95-stable overclock. If you can possibly hit anywhere near 80C core temp without Prime fail (or more than likely, the computer rebooting), I will retract my statement, because I believe you to be one of the few rare people who is as honest as I am.

 

[Jeff7181]

Originally posted by: myocardia
Jeff, I'm not going to spend 2 hours (I've got dialup, no choice) searching the internet for something to prove to you that a lower processor temperature yields a higher overclock, but why in the world do you think that spending ~$1000 on a Prometia yields higher overclocks every time than the same chip will run on air cooling? If you honestly think that I'm wrong about that, feel free to waste your time doing the research, but don't be surprised if you find out that I'm right about lower core temp equaling higher overclock. If you don't believe me, and don't want to do the research, unplug all of your fans, without lowering your overclock, then hold your hairdryer close to your cpu on high heat, while running Prime95 on a Prime95-stable overclock. If you can possibly hit anywhere near 80C core temp without Prime fail (or more than likely, the computer rebooting), I will retract my statement, because I believe you to be one of the few rare people who is as honest as I am.

You're still assuming I said that lower temperatures don't equal higher clock speeds... let me restate it again...

I am saying, a 5 degree celcius reduction in temperature is not enough to make a significant difference in the stability of the CPU or the ability of the CPU to be overclocked. By reducing the temperature 5 degrees celcius, you may be able to run a few megahertz faster, but it's not going to be a night and day difference as if you were using a Prometia setup that changes the temperature by over 50 degrees celcius. If you can get 200-300 Mhz more by decreasing the temperature by 50 degrees celcius, then decreasing the temperature 5 degrees celcius should get you 20-30 Mhz. Big forkin deal

 

[myocardia]

Originally posted by: Jeff7181
I'm saying people who recommend a better cooling solution at 50 degrees aren't correct. Bartons, Thoroughbreds, and Palomino's are all capable of running at 50 degrees C.

I agree completely, I would have to guess that most computers' processors in the world run ~50C under 100% load. And I also just noticed that you're on vacation, and in Florida. Get some for me, while you're there! (some sun and the other) Wait, isn't this Spring Break week? Don't tell me you're at Daytona Beach or Panama Beach...you lucky sucker!

 

[myocardia]

Originally posted by: Jeff7181You're still assuming I said that lower temperatures don't equal higher clock speeds... let me restate it again...

I am saying, a 5 degree celcius reduction in temperature is not enough to make a significant difference in the stability of the CPU or the ability of the CPU to be overclocked. By reducing the temperature 5 degrees celcius, you may be able to run a few megahertz faster, but it's not going to be a night and day difference as if you were using a Prometia setup that changes the temperature by over 50 degrees celcius. If you can get 200-300 Mhz more by decreasing the temperature by 50 degrees celcius, then decreasing the temperature 5 degrees celcius should get you 20-30 Mhz. Big forkin deal

Agreed, though I would think it would be in the neighborhood of twice that, since Prommy's reduce load temps closer to 90-100C.

 

[Jeff7181]

Ok, well a 90-100 C reduction in temperature gets you 200-300 Mhz... and water cooling which cools it down maybe as low as 25 at best assuming room temp is about 20. And you get 100-200 Mhz out of water cooling... do you see what I'm saying? AMD says an XP3200 will run 2.2 Ghz at 85 degrees C. It probably wouldn't run 2.4 Ghz at that temperature... and based on the water cooling example, to get the CPU to run 200 Mhz faster, I need to reduce the temp by 25 degrees C... so if we bring it down to 60 degrees C, it should run at 2.4 Ghz (and mine does). Now if I want to run another 200 Mhz faster, I need to reduce the temp by AT LEAST 25 C again... because as we know, this doesn't scale linearly, or we'd have 8.25 Ghz Bartons on Liquid Nitrogen. So what would decreasing the temp 5 degrees do? Well, 5 is 1/5 of 25... and 1/5 of 200 is 40... so by reducing the temp 5 degrees you'd gain about 40 Mhz... THIS is why I don't believe WhiteBoy's claim of reducing his temp 5 degrees and gaining 150 Mhz.

Oh and by the way... my vacation to Florida is a family vacation, we're visiting my grandpa at his condo on Marco Island (near Ft. Meyers and Naples)... his health is slowly turning downhill so I'll be spending most of my time with him cause there's no telling when the last time I'll see him will be. He's had 4 back surgeries and now has metal rods in his back, he's had two surgeries on his eyes, and he just had a pace maker put in a few months ago cause his heart rate slowed down to 40-50 beats per minute at times. So this next week will just be quality time with gramps

 

[myocardia]

Well, I don't know how old your grandfather is, but having a pacemaker won't reduce his longevity hardly any at all, especially since his bradycardia wasn't pronounced. As long as his back's better, don't be a bit surprised if he drags YOU down to the beach. He should have his spunk back, now that his heartrate is at a normal level again. Like clinical depression, most people don't realize that they have a problem with bradycardia until they go to the doctor/hospital for some other reason. They've just become used to it, and I would say that the vast majority of the patients I've had are amazed at how different (normal isn't normal to you unless you're used to it) they feel, especially concerning their energy levels. Uh, this isn't on-topic, is it? Anyway, you might ask him if he wants to take some lawn chairs down to the pier/dock and do some fishing or something slightly active. I'm sure it would be a high point for him, and quite possibly for you. See you later.

edit: I forgot 2 things: bradycardia is an abnormally low hearrate, and you may very well be right about the extra 150mhz, but I'm kind of tired of debating it for now. You should be having fun with your grandfather anyway, instead of debating with me!

 

[WobbleWobble]

Ok, this thread is long so I skimmed it briefly. If I say something that has already been said, feel free to slap me or somethign

Interesting you brought this up. I've been playing around with my XP2500+ mobile overclock these last few couple of days. I notice that heat at and above 50C socket on load does hurt my potential at overclocking. I am not able to run Prime95 at 2.4GHz with my socket temp at about 55C and diode at 65C. But if I kick up my fan speeds and lower my temps to about 40C and about 50C diode, it'll pass Prime95. Both temps are under the thermal limits, but the overclock is hindered.

For a stock speed CPU, what you say may be true. I have older non-overclocked AMD CPUs running at 50C+ and they're Prime95 stable. But this is what I'm experience right now and I'm actually doing tests that counter what you're saying.

So my recommendation is, if your CPU is running 50+ at stock speeds, you're fine. If you want a higher overclock, you should look into better cooling.

 

[Wuzup101]

mmm... i have a 2100+ @ 2.2ghz at home load is around 57C- 60C I would venture to guess w/ the stock hsf from my 2500+ barton on there. Prime stable for 24hrs. Heat does affect an overclock, but definitely not as much as everyone thinks. Watercooling and phase change get several hundred mhz on occasion, but you have to remember they are significantly lowering the cpu temp. You aren't doing this by upgrading to a great hsf over an average hsf.

If you think about it it's really insane what we do. We (yes myself included) typically have more money in cooling then we do in the damn processor in our computers.

Anyway, I have to waste the money somewhere - mise well be into my computer instead of drugs and alcohol.