Whats an allowable temp for a CPU?

[busmaster11]

Got a mobile Barton here... Gets 55 degrees after a night of Folding@home... Is that bad?

What's bad?

 

[Cheetah8799]

not bad, but high. WIth poor ventilation and a stock heatsink, that's about normal during 100% cpu usage.

Most people here will say it's insane high, but it isn't really. I think the specs for those cpus (I have one) is 85C before it will die...

You may want to try to lower it down to around 40C - 45C at 100% load if possible.

Are you overclocking it at all yet?

 

[busmaster11]

Originally posted by: Cheetah8799
not bad, but high. WIth poor ventilation and a stock heatsink, that's about normal during 100% cpu usage.

Most people here will say it's insane high, but it isn't really. I think the specs for those cpus (I have one) is 85C before it will die...

You may want to try to lower it down to around 40C - 45C at 100% load if possible.

Are you overclocking it at all yet?

Wow...

It's a 2500+ Barton runing at 2200mhz (a 2-300mhz overclock) and 1.6 volts... It's using pretty stock cooling and a full tower Chieftec AL case with about four case fans...

Do you think fancy Hs compound makes a difference?

 

[WobbleWobble]

Yes, fancy compound makes a difference. Try AS5 and you should notice a nice little drop.

That temp is high for 1.6V, unless you're reading the diode temp. Then that sounds about right. If it's a socket temp, then it should be lower. Even the a cheap cooler should be able to do better than that.

85C is the highest a Barton should go. But that's not 85C socket temp, but internal temperatures. So keep it below 70C socket to be safe. Your temps aren't dangerous, but they are high.

 

[Jeff7181]

That's a perfectly fine temp... I don't consider it high. Mine's running 53 C right now while encoding a movie with divx. BTW... this is my mobile Barton at 2.4 Ghz (209x11.5)

*EDIT* I just let my friend use some of my AS5 on his Pentium 4 2.0 @ 2.3... it's the retail heatsink with that black thermal crap (which by the way isn't as hard to get off as people say)... AS5 reduced the temps by about 5 degrees C

 

[pelikan]

I think that software temp readings are not accurate or comparable between different motherboards (or sometimes even the same motherboard with different bios revisions). If it will run prime95 overnight then the temps are fine. Prime95 will produce errors if the cpu is too hot.

 

[busmaster11]

Originally posted by: WobbleWobble
Yes, fancy compound makes a difference. Try AS5 and you should notice a nice little drop.

That temp is high for 1.6V, unless you're reading the diode temp. Then that sounds about right. If it's a socket temp, then it should be lower. Even the a cheap cooler should be able to do better than that.

85C is the highest a Barton should go. But that's not 85C socket temp, but internal temperatures. So keep it below 70C socket to be safe. Your temps aren't dangerous, but they are high.

How can I tell? I'm running Speedfan 4.11 on a Shuttle Nforce2 board and it tells me my temp2 value is 48 degrees running F@H... I'm assuming that's my CPU temp because its the highest reading available...

 

[WobbleWobble]

AIDA32 will read both the socket and diode temps.

 

[coachman]

Originally posted by: WobbleWobble
AIDA32 will read both the socket and diode temps.

Where in AIDA32 is the selection for Socket/Diode temp?

 

[im2good4u]

hey is mine fine?
P4 2.8C @ 3.5GHZ stock cooling
P4P800-DELUXE
idle: 37C
loaded: 48-50C
or do i have to add some extra cooling?

 

[WobbleWobble]

Originally posted by: coachman

Originally posted by: WobbleWobble
AIDA32 will read both the socket and diode temps.

Where in AIDA32 is the selection for Socket/Diode temp?

Computer > Sensor

 

[Shaocaholica]

From what I can tell, your mobo will either report a sensor or diode temp but not both. You should only have 1 CPU reading and shouldn't be worried about switching between the two since you will most likely not have a choice.

 

[Jeff7181]

I thought all nForce2 boards could read the diode temp AND socket temp. Mine does anyway, but it's an A7N8X Deluxe.

 

[WobbleWobble]

My board reports both. Athlons that are XP and newer can read it. The Thunderbird models and older couldn't.

 

[kin3]

I have an A7N8X Deluxe. MBM says socket 50, diode 19, and case 19. Is this ok?

 

[XxervioCSO]

45-50 is really to high, of course you know what they say the cooler the better.

i have athlon 2500 running 220x10 at 1.7v and full load it only goes to 38 degrees. i use a valcano 7+ that i got when i purchased the processor. it's semi loud but cools great.

 

[XxervioCSO]

 

[Jeff7181]

Originally posted by: kin3
I have an A7N8X Deluxe. MBM says socket 50, diode 19, and case 19. Is this ok?

You have your diode temp setup wrong... you set it to use the case temp sensor. Your diode temp will usually be about 5 C higher than your socket temp.

 

[Jeff7181]

Originally posted by: XxervioCSO
45-50 is really to high, of course you know what they say the cooler the better.

i have athlon 2500 running 220x10 at 1.7v and full load it only goes to 38 degrees. i use a valcano 7+ that i got when i purchased the processor. it's semi loud but cools great.

Nope, it's not too high... and your full load temp is not 38 degrees with air cooling... don't care what your readings tell you, they're wrong... it's impossible, especially with a volcano 7.

 

[WobbleWobble]

My A7N8X-Deluxe socket temp is ~10C less than diode temp under full load. When it's not under full load, the difference fluctuates too greatly to say.

 

[XxervioCSO]

i get those readings from the winbound software that came with the abit nf7-s. also stats case temp to be 19-20 degrees. if winbound is wrong and sisoft sandra is wrong then let me know a program that will read right.

 

[Jeff7181]

Originally posted by: WobbleWobble
My A7N8X-Deluxe socket temp is ~10C less than diode temp under full load. When it's not under full load, the difference fluctuates too greatly to say.

I've found that to be a sign of either poor thermal conductivity between the core and heatsink (either because of poor contact with the core or a low quality thermal compound), or a low quality heatsink. A third possibility would be that the socket temp sensor reads on the low side, which is very likely, probably more likely than the other two.

 

[Jeff7181]

Originally posted by: XxervioCSO
i get those readings from the winbound software that came with the abit nf7-s. also stats case temp to be 19-20 degrees. if winbound is wrong and sisoft sandra is wrong then let me know a program that will read right.

The software reads those temps from the the motherboard's sensors... if the motherboards sensors are calibrated incorrectly, any software will read the incorrect number.

Shut your computer off for a couple hours... when you turn it back on IMMEDIATELY enter the BIOS and see what the case temp says. If it says something like 10 or 15, then you know it's reporting 5-10 degrees lower than the actual temp, and I'm not sure about other software, but Motherboard Monitor allows you to calibrate the software to make up for the inaccuracy of the sensors/motherboard. Don't pay attention to the CPU temp it reports because the CPU temp jumps 10-15 degrees within 5 seconds after you turn the computer on... case temp on the other hand takes a few minutes to get up to the normal temperature.

 

[WobbleWobble]

I've found that to be a sign of either poor thermal conductivity between the core and heatsink (either because of poor contact with the core or a low quality thermal compound), or a low quality heatsink. A third possibility would be that the socket temp sensor reads on the low side, which is very likely, probably more likely than the other two.

I'm using Arctic Silver 5 with a Zalman CNPS-7000A. I don't think there's any problem with the mounting of the heatsink. I just think that the sensors are doing the job and that's why we have both

 

[Jeff7181]

Originally posted by: WobbleWobble

I've found that to be a sign of either poor thermal conductivity between the core and heatsink (either because of poor contact with the core or a low quality thermal compound), or a low quality heatsink. A third possibility would be that the socket temp sensor reads on the low side, which is very likely, probably more likely than the other two.

I'm using Arctic Silver 5 with a Zalman CNPS-7000A. I don't think there's any problem with the mounting of the heatsink. I just think that the sensors are doing the job and that's why we have both

Just curious... how long was it under load before you noticed a 10 degrees C difference? Diode temp jumps immediately, and the socket temp can take 10 minutes or more to be saturated with heat and finally stabilize.