(load_temp - idle_temp) > 5 degrees? It's not all about getting the lowest temp.

[propellerhead]

Is your CPU temp under load very different from the temp at idle?

My CPU temp under load is 40-41. When idling, it sits at 39-40. To me, that's good because the CPU does not go through a heat/cool cycle all the time.

Anyone care to explain why some some people have large variances between the idle and load temperatures and some don't? Is this a good thing or a bad thing?


===================
Epox 8KHA+
Athlon XP 1600+
Volcano 6Cu+ (7000 rpm)

 

[Mikewarrior2]

well, a lto of it depends on how you're measuring temps... on a stock amd setup, with socket-thermistor readings, since these readings are often compressed 2X or more, minimal idle ot full load changes will exist.

Looking at internal diode readings, you see this: AthlonXP internal diode grease test, use it compare idle versus load temps.

Even with watercooling, you've got usually a 10-15C difference between load and idle... with one of hte premier air cooling setups, you see 20-25C differences.... soemthing to consider, but its fairly normal for the large differences when measured at the core level, since load to idle and back do not take long to change temps.


Mike

 

[Match]

Besides what Mikewarrior2 mentioned above, I can think of two other reasons why people get such varied differences between idle and load temps.

1) It depends on how good your cooling setup is to beging with. If you have a good setup, idle temps will be close to ambient and load temps won't be too far off from idle, so all the temperatures will be relatively close together. If you have a poor setup, idle temps can be well above ambient and load temps even higher, so the temperatures will be relatively far apart. So the difference between idle and load on a good setup is less than that on a poor setup. Make sense?

2) "load" is a relative word. Even though your processor may be working at 100% it doesn't mean it's as hot as it can be. It depends on how you're loading it. For example, when I run Seti on my computer I get a lower temp than when I use CPUBurn (which is designed to heat up the processor). Here are some of my temps on a 1.2 GHz Athlon (using the ever-inaccurate socket thermistor readings):
system 26
idle 43
SETI 47
CPUBurn 53

 

[propellerhead]

<< with one of hte premier air cooling setups, you see 20-25C differences >>



Won't this cause fatigue from repeated heating and cooling? I don't think I want to spend the money on one of those "premier air cooling setups" if my CPU temp will fluctuate from maybe 35 degrees to 50 degrees.




<< If you have a good setup, idle temps will be close to ambient and load temps won't be too far off from idle, so all the temperatures will be relatively close together. >>



I agree with you on that, Match. Effective cooling should be based on the HSF's ability to keep the CPU temp low AND stable with very little variance.

 

[Mikewarrior2]

Propellerhead,

The big variance is probably there, whether or not the socket-thermsitor picks it up or not.... its just the way CPUs are when temps are measured internally....

You'd see the same probably wtih a cheapy heatsink... the fluctuation would just be bigger, not smaller.



Mike

 

[Jman13]

My CPU temp on my AthlonXP 1700+ @1540 is 32C idle, 45-47C during SETI. Idle temps go down considerably (10C) when you have done the VIA Halt instruction fix...at no cost to performance.

Jman

 

[propellerhead]

Mikewarrior2: Your posts remind me of this joke about engineers. The guy in the air balloon is lost and yells at some guy in a building and asks where he was. The guy in the building tells him that he is 25 feet off the ground and 25 feet north of the building, or something like that. Then the guy in the balloon tells the guy in the building that he must be an engineer because his answers are true and correct, but doesn't really get him anywhere. But I do appreciate your posts, even if they are a little off subject.

Going back, I don't think it really matters much whether the CPU temp is measured from the top or bottom or side or inside or outside or upsidedown or rightsideup or forwards or backwards, BTU, Kelvin, Celsius or Fahrenheit . I was talking about the change in temperature. I'm only interested in the difference between idle and load temps. And as long as the method of measuring the temperature remains constant, we should be ok.

Picture this really simple scenario... In most cases, you come home from work or school and you turn on your PC. The temp reads somewhere around room temperature. You let your PC sit idle, usually at the Windows Desktop, and you have your idle_temp. Then run SiSoft Sandra's Burn In, or play HalfLife, or fly around in MS Flight Simulator, or anything that you define as load. At that point, you have your load_temp.

So, don't you think the ideal situation would be when the room_temp and idle_temp and load_temp are pretty close to each other? Isn't that better than saying my idle_temp is 25 deg C but the load_temp is like 45 deg C?

Anyone else have thoughts on this?

 

[Mikewarrior2]

Propellerhead,

That's entirely the problem... how CPU temps are shown is entirely dependant on where they are measured.... Even across different motherboards, all with socket-thermistors, there is a significant difference in temps read between them. The temp difference between idle nad load, on the same motherboard, are different from each measurement type.

Yes, ideally you want temps that are close to each other... however, with certain motherboards and setups, that isn't possible... and that certianly isn't possible with internal diode temp measurements.


Mike

 

[propellerhead]

Mike:

You're really stuck on this where and how to measure temps. So, tell me, why are temps different on different motherboards?

 

[Mikewarrior2]

i'm stuck on it because its absolutely the truth... there's no way around it...

the reason temps are different across different motherboards? Because of various factors, including socket-thermistor location (yes, even slight deviances result in different temps) and different motherboard manufacturer compensation techniques). Hence why the idle to load "norms" are different between different motherboards.



Mike

 

[propellerhead]

Mike:

So what you're saying is that different mobos read temperatures differently? That's great! I'm glad you pointed that out. In the two-hundred-thousand-million articles out there on the web, none of them covered the fact that different motherboards will register different temperatures as thoroughly as you did. You are da man! You rule! You are the most-knowing. Great! I am now aware of that fact that my mobo will read different temps compared to your mobo. Everyone else catch that? Great! Everyone now knows that temps are different across different motherboards. Cool. I have it on a Post It note on my monitor. Different mobos read temperatures differently. Different mobos read temperatures differently. Different mobos read temperatures differently. Awesome!

Ok. So let's try something different. When I put my CPU/mobo together, the mobo monitor would register x deg C when it just sits at the Win98 desktop. But after running HalfLife for 30 minutes, it would register y deg C. There was a 12 degree difference between x and y.

I wasn't comfortable with that so I replaced the stock HSF with a Volcano Cu+. So then, the mobo would register p at the Windows desktop and q after playing 30 minutes of HalfLife. There was a 2 degree difference.

I put the original HSF on and got the same 12 deg difference. I put the Volcano back on and got the same 2 deg difference. Same mobo. Same CPU. Same case. Same room. Same location within the room. Same room temperature. Same desk. Same chair. Same monitor. Same hard drive. Same mouse. Same everything else I can think of.

Can anyone explain why different set ups on the same mobo could have big differences between idle and load temps and some don't?