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

[myocardia]

Originally posted by: 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.

This was going to be my reply to Jeff, until I decided to forego the debate: I am under the impression that, just like how far it will go with overclocking at any particular vcore, I'm convinced that each and every processor has it's own temp limit, that will vary somewhat from cpu to cpu, even if both are the same speed of the same chip (2500 Barton's, or 1700 T-bred B's, etc.) I have no way of proving this, but it doesn't mean that it isn't true, or at least quite possible. For instance, it seems that T-bred B's handle heat better than Barton's, but how could there be such a huge discrepancy between all of the 2500 Bartons, if we just forget about every other Athlon ever made? Because they aren't exactly identical, that's why some will OC past 11x200 at stock voltage, but others (not alot, I give you detractors that) can't seem to make it all the way to 200fsb, no matter what the voltage? I personally believe that the two are related, but even more than that, I believe that they prove each other.

 

[AWhackWhiteBoy]

Originally posted by: Jeff7181

Show me.

i can't,i never saved the website and i can't find it again on google, however i can say in my experience it is definatly true. at 2ghz my heatsink will run the CPU at room temp, soon as i start pushing 2.2 and 2.3 the temp shoots up. try for yourself.

 

[Acanthus]

Originally posted by: AWhackWhiteBoy

Originally posted by: Jeff7181

Show me.

i can't,i never saved the website and i can't find it again on google, however i can say in my experience it is definatly true. at 2ghz my heatsink will run the CPU at room temp, soon as i start pushing 2.2 and 2.3 the temp shoots up. try for yourself.

it would depend on the stepping of the chip. 5 degrees C cant increase your OC by that amount.

 

[AWhackWhiteBoy]

Originally posted by: Acanthus


it would depend on the stepping of the chip. 5 degrees C cant increase your OC by that amount.

tell that to my setup,thermal diodes aren't all that trustworthy to begin with.

 

[myocardia]

I'm not trying to defend Whack, but it's quite possible that he made more than one change at the same time that he installed his new heatsink, netting him an extra ~150mhz. Let me give an example, my XP2400. It will go higher with a higher multi. Now please don't tell me that multixfsb gives you your frequency, so of course it goes higher. I figured that out with my first OC- a 486DX! Anyway, I don't call it a successful overclock unless it's Prime95 stable, but with exactly the same cooling, it's only Prime95 stable where I run it at: 2238mhz/10.5x212.58. I run it at that because it provides the highest performance. Most of the stuff that I run benefits most from a higher fsb. But, I can say without lying that my processor is totally Prime stable at 2334mhz, because it's Prime stable at 11.5x203. Yes, it does require a .025v vcore bump, but what if he bumped his vcore a little when he installed the heatsink, thinking the new heatsink would help with the extra heat produced? I have an absolutely horrible memory now, but even an average person with a "normal" memory can't remember everything. Anyway, he sure does seem adamant about something that he could have just never replied to, and we all would have forgotten about it sooner or later.

 

[Jeff7181]

WobbleWobble... again, your temp was 65 and unstable... if you read my original post, I said 50 degrees is not too high... 65... yeah, maybe a bit too high (but still within AMD specifications, at stock speed, like you said), so you reduced it by 15 degrees and now you can run 2.4 Ghz. By White Boy's claims, you should be running 2.65 Ghz right now. Since AMD rates them at 85 degrees at stock speeds, and you had to cool it to 50 degrees to get a 200 Mhz overclock, that's more realistic. Your situation is perfectly believable... 5 degrees getting you 150 Mhz is not.

AWhackWhiteBoy... I don't think you're reading the CPU's diode... I think you're reading socket temps. And don't worry about the chart, I don't need to see it.

Myocardia... I believe he was diagnosed with "heart block" ... I looked it up a while ago, and basically it's just the electrical impulses that make the heart beat are being "blocked" or somehow interrupted... his pacemaker pretty much solved that problem... but he's 80 years old... surgeries are hard on him at that age... his back still bothers him, it's difficult for him to get around, can't bend over very far cause 4 of his lower vertibrae (sp?) are fused with a metal rod attatched to them as well... his doctor wanted to do another surgery and he figured enough was enough, no more surgeries.

**EDIT** Ok, one more reply to Myocardia's reply... that's very much possible... a voltage increase would explain the increase in speed. BUT... that's not due to a 5 degree reduction in heat, it's due to an increase in voltage... so again... I still don't believe 5 degrees C netted him 150 Mhz.

*EDIT* No more replies from me until I arrive in Florida... about to start the 22 hour drive... :Q

 

[AWhackWhiteBoy]

but it's quite possible that he made more than one change at the same time that he installed his new heatsink, netting him an extra ~150mhz

went from thermal grease and a $15 heatsink to a $50 heatsink with arctic silver 5 and 130CFM rear fan. it still only yeilded 5C because i refuse to run the fan at full speed.

AWhackWhiteBoy... I don't think you're reading the CPU's diode... I think you're reading socket temps. And don't worry about the chart, I don't need to see it.

yes,i'm reading the socket's temp because the diode changes too much, i consider the socket a rough "mean".

 

[WobbleWobble]

Jeff, my point is that even though the temperatures are still lower than AMD's max thermal rating, it can hinder overclocks. The feeling I got from your posts is that as long as it's under AMD's rating, it'll run fine.

Also, we have a situation with the socket and diode temps. I would think that most uses report their socket temps as that's usually what's given out. Which is why I also gave you my diode temps, I could have given you my socket temps alone and then my temps wouldn't look that bad. But the diode temps can be considerably higher than the socket temp.

On the side note, 22 hours drive?! I feel sorry for your ass

 

[myocardia]

Wobble, the diode temps are nearly always right at a 10° higher difference from the socket temps. I personally just go by the socket temp, since I know that the diode is going to be between 9-11° higher. As long as everyone is using the same, it works pretty well, I think.

 

[Wuzup101]

About that 85C - remember that that temp is a maximum operating temperature. AMD doesn't intend for their processors to be run at that high of a temperature. They just say that they shouldn't be harmed by temps under 85C. I'm sure if their processors were perfectly happy running at that speed they woldn't make stock heatsinks that keep them well below 85C.

 

[wicktron]

50C IS too high. If I could cool CPU down to say 40C, I could push about 100-200 more MHz out of the chip. So YES, it DOES matter if temps are that high. It may not be high if you are content with your overclock, but when trying to squeeze out every last MHz, it matters an extremely great deal. Watercooling/Phase Change would be almost unecessary if staying between 50C-70C was "acceptable."

 

[SpideyCU]

wicktron, you're mixing words. I know what Jeff7181 is trying to say, and people keep misinterpreting it. Because people keep saying "50 C is too high!" when they're thinking in terms of "I'm not maximizing my overclock as much as I could", OTHER PEOPLE reading this are interpreting it as "50 C is too hot and your CPU won't run stable", which is just wrong. He's trying to clear the air, which I give him props for because it's very easy to start confusing folks here.

Are you even reading what Jeff7181 wrote?

This may not be the thread for it, but since we're discussing 5 C drops with this cooler over that, I have a slightly different question: instead of discussing just CPU temps when gloating about their numbers, why don't people mention their socket temperature (reportedly) as compared with their CASE temp? Quoting absolute numbers means nearly nothing when discussing the effectiveness of a heatsink. I mean, for example, my CPU socket temp is usually 15 C above the case temp, 20 C under load. Though, I could've sworn that when I installed this Thermalright HS a few months ago, it was 10 idle/15 load. :disgust:

 

[pamf]

ive got an aqxea 2500+ @ 2.3ghz/400fsb running 36c under full load (well, seti anyhow) with an slk-800u and some as5..

ive also got an aquca 2500+ running stock around 55c with moderate load, with the stock cooler and some as5..

the aquca WAS running around 70c most of the time, but this was with some as3, an older stock cooler (ive got a newer one with a copper base now), and poor case ventilation..

too hot for my tastes, but still fine..

70c was too hot though. when the case is hot to the touch, its time to buy some extra fans..

 

[DAPUNISHER]

Jeff, this cold one is for you brudda :beer: It's tiresome to continuously see people adhering to the old rule of thumb about AMD CPU temps that came about because of the more inaccurate socket temp reading that was relied upon back then. With usually more accurate on-die readings a temp reading in the low 60's is fine if stability isn't effected. I realize there are still some reading inaccuracies but the days of a system reading 50c CPU temp and being unstable because of overheating are largely behind us.

 

[Atlantean]

Originally posted by: 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.

Mine was only able to go up to 2.1 ghz, I changed to better cooling and as5, and it dropped the temp by 10 degrees so I put it up to 2.2 ghz and it was stable. Previously when I had been at 2.2 it would crash occasionally and wasn't as stable as it should have been and I blame the problem on heat cause it doesn't do it now.

 

[Jeff7181]

Thank you SpideyCU and DAPUNISHER.
I think some people here just like to contradict others, and that's fine, it opens up the discussion which brings up some very valid and useful information.
As I've said before, I have no doubt lowering the temp will allow higher overclocks, but the point of this post was exactly what SpideyCU pointed out... people are giving bad advice... 50 degrees C is not too hot. For an extreme overclock, sure it is... for an EXTREME overclock, 20 degrees C is too hot. But I see people who are asking questions, and they mention their CPU temp is 50 degrees C, then someone tells them that's too hot and they need to do something about it, which simply isn't true, unless like others have pointed out, you're squeezing every last MHz out of the CPU, in which case, you shouldn't have to be told lower temps are better... and you SHOULD have understood my point =) If you'll notice in the original post, I mentioned that AMD ok's the processors up to 85 degrees C at stock speeds.

*EDIT* By the way... hello from Marco Island, FL!

 

[John]

Originally posted by: Jeff7181
50 degrees C is not too hot.

That is a true statement. I've been an avid overclocker since 1998, and since I live in West Texas the temps can get extremely hot out here. Over the years I've always seen people ranting and raving about their low cpu temps, and often wondered if they weren't BS'ing so they "fit in". My current mobile 2500+ @ 2500MHz 1.76v cooled w/ a SLK800a & 80mm Panaflo L1A runs 51C with no load and 61C under full prime95 Torture Test load. Even when it was Winter and the outside temps were 32F, I could not overclock my cpu's any higher. As it's been pointed out, people seem to put too much emphasis on low cpu temps.

Anything under 65C load is acceptable.

 

[Jeff7181]

I think it should also be pointed out that different motherboards measure "socket temps" at difference places. However, diode temps are taken right from the thermal diode in the CPU core. THAT is why I almost always give my diode temps rather than socket temps. Socket temps vary too greatly from motherboard to motherboard so there's no real basis for comparison. Not to mention that some just report the wrong temps to begin with. The best way to check that is to use the case temp and a thermometer to find out how far off the motherboard is reporting the case temp, then adjust your monitoring software to compensate for the amount that it's off by, and apply that same amount to the socket temp as well.

 

[Jeff7]

Main system (Tbred 2100@2.2GHz)
50C CPU avg; 45 - 53 range
33C Case avg; 29 - 35 range

Secondary system (Tbred 1700@1.9GHz)
53C CPU avg; 50 - 56 range
35C Case avg; 34 - 37 range


A bit on the warm side, but they both seem to be quite stable.
The room temp right now is 28C, which is also on the warm side. But then, it's a small, insulated room, and these two PC's act as effective space heaters.
Both systems run almost continuously too.

Those are socket temps too; the motherboards (NF7-S and 8RDA+ respectively) can't seem to give me diode readings. Do the Tbreds simply not have the diode? Is that a Barton thing?

 

[TerryMathews]

Originally posted by: AWhackWhiteBoy
if you look at the heat output of the bartons on a chart you'll see at 2.2ghz it curves up exponetially.

Physically impossible. You would see a continually rising increase in temperature if your processor was generating more heat than your fan was able to remove from the heatsink, but it can't be exponential. Violates the laws of nature.

 

[Sokratz]

I was actually just coming to the forums to ask about this. I got most of my new parts (nf7-sv2 mob,xp2500) in but not the cooling stuff. I'm using my old hsk which I think is fairly junk. Anyways, using the winbond/mbm montors (for the life of me i can't get abiteq to work, either errors w/"open file error" or "bad namespace" wtf?) it'll show my temps of 37c idle, 45 load @ 200x11.5@1.7 which seems pretty good to me. Even if there were a 20% difference from diode temp. However, if I have in the bios to shutdown the computer at say 75C, damn thing will reboot on me all the time (even while in BIOS-eep). This makes me feel "funny".

feedback?

Thanks,
Sokratz

 

[Jeff7181]

Originally posted by: Sokratz
I was actually just coming to the forums to ask about this. I got most of my new parts (nf7-sv2 mob,xp2500) in but not the cooling stuff. I'm using my old hsk which I think is fairly junk. Anyways, using the winbond/mbm montors (for the life of me i can't get abiteq to work, either errors w/"open file error" or "bad namespace" wtf?) it'll show my temps of 37c idle, 45 load @ 200x11.5@1.7 which seems pretty good to me. Even if there were a 20% difference from diode temp. However, if I have in the bios to shutdown the computer at say 75C, damn thing will reboot on me all the time (even while in BIOS-eep). This makes me feel "funny".

feedback?

Thanks,
Sokratz

Maybe check the installation of the heatsink. Remember, as someone pointed out, the socket temp is rarely indicate that the heatsink isn't making proper contact... at idle it might look fine, under load it might look fine, but only because heat isn't being transferred from the core because of poor contact with the heatsink. So, first thing I'd do is check the heatsink installation. You are using some sort of thermal compound, correct?

 

[pelikan]

Our software temp monitoring is not necessarily accurate, no matter if its read off the die or socket. Most people know that 45C in one system can = 60C in another.

If a cpu will run prime95 for 8 hours or more then the temps are fine. I was glad to see the title of this thread because I think people need to stop worrying about their temps and just confirm stability.

Lower temps have definately increased my overclock. But not by much. I only got 100 MHz going from high end air cooling to watercooling with 10-15C water. Its a good thing that I switched to water mostly for the silence and modding or I would have been disappointed. It seems to me that it takes sub-zero temps to make an overclock really skyrocket.

 

[Jeff7181]

It seems to me that it takes sub-zero temps to make an overclock really skyrocket.

True, because at these low temperatures, the physical properties of the CPU change, allowing clearer electrical signals and faster switching. But when you're talking about the difference between 50 and 60 degrees C, there's almost no difference. That was Kryotech's claim back when they were still building their "thermally accelerated" computers.

 

[AWhackWhiteBoy]

Originally posted by: TerryMathews

Originally posted by: AWhackWhiteBoy
if you look at the heat output of the bartons on a chart you'll see at 2.2ghz it curves up exponetially.

Physically impossible. You would see a continually rising increase in temperature if your processor was generating more heat than your fan was able to remove from the heatsink, but it can't be exponential. Violates the laws of nature.

how do you figure? anything above 2.2ghz is out of spec and its probably a boost in the voltage in order to do it, hence the huge jump.