Weird Temp Issue, please help.

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Problem explained below. Please let me know what you think. First I will give you my temps and other stats of all my OC attempts.



OC attempts. Stressed using dual prime 95's
Voltage is at 1.3(as reported in Speedfan and PC Wizard. I have it set at 1.325 in BIOS, showing actual of 1.3 in BIOS

E6600
eVGA 680i A1
G.skill DDR800 HZ's
eVGA 8800GTS 640mb




2700Ghz(300x9)

Program/temps(idle/load)

TAT - 35/30(cores idle) - 45/42(cores load)
Speed Fan - 41(idle CPU Temp), 53(load CPU Temp)
PC Wizard - 34/29(Cores idle) - 46/43(Cores load)
CoreTemp - 34/30(Cores idle) - 46/42(Cores load)
nvMonitor - 40(idle)/53(load)
BIOS. - 43(idle)

2835Ghz

Program/temps(idle/load)
- Dual Prime 95's
TAT - 36/31(cores idle) - 46/43(cores load)
Speed Fan - 41(idle CPU Temp), 55(load CPU Temp)
PC Wizard - 34/30(Cores idle) - 47/43(Cores load)
CoreTemp - 34/30(Cores idle) - 46/43(Cores load)
nvMonitor - 40(idle)/54(load)
BIOS. - 43(idle)

2996Ghz

Program/temps(idle/load) Dual Prime 95's
TAT - 36/31(cores idle) - 47/44(cores load)
Speed Fan - 41(idle CPU Temp), 55(load CPU Temp)
PC Wizard - 34/30(Cores idle) - 47/45(Cores load)
CoreTemp - 34/30(Cores idle) - 48/45(Cores load)
nvMonitor - 40(idle)/55(load)
BIOS. - 43(idle)

Orthos(1hour)-
Program/temps(load)
TAT - 48/44(cores load)
Speed Fan - 56(load CPU Temp)
PC Wizard -48/45(Cores load)
CoreTemp - 48/45(Cores load)
nvMonitor - 56(load)


TAT
Program/temps(load)
TAT - 53/50(cores load)
Speed Fan - 57(load CPU Temp)
PC Wizard - 53/50(Cores load)
CoreTemp - 53/50(Cores load)
nvMonitor - 56(load)

-------------------------------------------------------------------------------------
Here is where the problem happened. Not sure what to think of it. Windows was running stable. Idle temps where even lower than previous testing because I stuck a big Fan in front of the PC(ambient temps are starting to warm up here). I watched the dual Prime 95's run for about 15 minutes(Never got above 54 degrees C before I left) and then went to go get some food. Came back about 20 or so minutes later and it was still running and my CPU temps in nvMonitor, SpeedFan, as well PC Wizard said 102. The PC wasn't locked up, Prime 95 was still running(no errors, ect). I shut down prime 95 and the temp stayed at 99 degrees. I turned it off and booted into BIOS and it said that I was at 40 degrees idle.

I booted back into windows at the same 3109Ghz and all my temps are idle at about 37 degrees for the CPU.


3109Ghz(345FSB)

Program/temps(idle/load) - Dual Prime 95's
TAT - 32/27(cores idle)
Speed Fan - 37(idle CPU Temp), 102(load CPU Temp) <-----WTF???
PC Wizard - 30/27(Cores idle)
CoreTemp - 30/25(Cores idle)
nvMonitor - 37(idle)/102(load)<-----WTF???

I would think I would have seriously damaged something if my temps really got to 102 degrees C. Especially for any length of time. But everything seems to be fine, like I said Windows wasn't even locked up. I also don't have Thermal control disable in BIOS.



I also realize that my core temps should be showing higher than my CPU temps. I am not sure why they are showing lower. I don't have core temps for speedfan because I jacking around with the offsets and I am not sure if I set them right so I didn't see a need in reporting them.



Thanks for your help.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
running Orthos on it now and it's stable (cross fingers) and at 51 degrees C. at full load. Who knows. I just don't see how the 102 degrees could have been right, maybe I was seeing things or it was a bogus temp reading even though ever monitoring program I had open was showing the same temp.


/shrug.

Hopefully I didn't screw it up. I figured I was pretty safe considering I have it at almost stock voltage.
 

little_scrapper

Distinguished
Aug 2, 2006
621
0
18,980
Turn on throttling. See what happens when temps get to 102.

My advice is get rid of the programs that say your temps are 102 (C im assuming)

Throttling should kick in between 83-85C is you really want to test it as I did.

I am curious what cooling your using to only have a delta of 10C. This doesnt seem right to me. Unless your using water and you folding taking place for idle temps.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Turn on throttling. See what happens when temps get to 102.

My advice is get rid of the programs that say your temps are 102 (C im assuming)

Throttling should kick in between 83-85C is you really want to test it as I did.

I am curious what cooling your using to only have a delta of 10C. This doesnt seem right to me. Unless your using water and you folding taking place for idle temps.

I have a Scythe Infinity.

It happened again with at 3.2Ghz. I was running Orthos for about 40 minutes with temps being around 55-57 degrees C. and all of a sudden they jumped to 98 degrees C. Again, no lock up or anything. Orthos was still running, and i was able to click start-shutdown.

SpeedFan 4.32, PC Wizard, and the nVidia monitor all said about the same high CPU temp.


I never disabled Thermal control in my BIOS so even with out stepping enabled it should have shut it down at that temp right? I have speed step, and C1E disabled, but I didn't disable thermal control like a lot of the OCing guides said. Could that be the problem? I don't understand why it would go from stable 56 degrees C to 100c in a matter of seconds(or less). Orthos was working on it's 2nd or 3rd pass(I think 3rd).

Idle temps right now at 3.204Ghz at 1.35v are (CPU - 37c in speedfan, PC Wizard, and NVidia's nvMonitor)
 

noobie123

Distinguished
May 11, 2007
7
0
18,510
I have this same problem on a p5b deluxe and e6600 but I would think its some kind of software error/bug because if you look at core temps, they don't have a large spike like the cpu temp. (cpu temp should be roughly 5 degrees below core temp so it makes no sense for cpu temp to jump like 60 degress higher than core temp)


I'm no expert but i think speed fan gets its readings from the manufacturer's monitoring software, which gives bogus values during stress testing all the time.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
I have this same problem on a p5b deluxe and e6600 but I would think its some kind of software error/bug because if you look at core temps, they don't have a large spike like the cpu temp. (cpu temp should be roughly 5 degrees below core temp so it makes no sense for cpu temp to jump like 60 degress higher than core temp)


I'm no expert but i think speed fan gets its readings from the manufacturer's monitoring software, which gives bogus values during stress testing all the time.


I think you are right because my core temps don't move at all. Just my CPU Temps jump from like 55c to 100c in a matter of a second or two. It doesn't make since.
 

little_scrapper

Distinguished
Aug 2, 2006
621
0
18,980
Keep in mind that there is NO WAY to read the "actual" C2D temp via software. It doesnt have the capability. ALL programs are basically starting with a basline temp and using some assumed values, like Tj Max of 85C, and figuring the temp from there. Guessing, if you will, by making calculations using assumed numbers.

This is something that you MUST keep in mind.

As for your 60C jump....not likely. Do ALL programs show this jump?
or just some of them?

If possible find a friend to drop your CPU into his board.
As for which to RMA if it comes to that. My "GUESS" would be to RMA the mobo. But that a guess at this point.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Keep in mind that there is NO WAY to read the "actual" C2D temp via software. It doesnt have the capability. ALL programs are basically starting with a basline temp and using some assumed values, like Tj Max of 85C, and figuring the temp from there. Guessing, if you will, by making calculations using assumed numbers.

This is something that you MUST keep in mind.

As for your 60C jump....not likely. Do ALL programs show this jump?
or just some of them?

If possible find a friend to drop your CPU into his board.
As for which to RMA if it comes to that. My "GUESS" would be to RMA the mobo. But that a guess at this point.

It showed that on more than one program. SpeedFan 4.32, PC Wizard, and the nVidia monitor all said about the same high CPU temp. Core temps didn't jump. If this really was a heat issue I would think that they would have jumped up as well.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Anybody have a suggestion on why it would do this? Is it just misreading my temp all of a sudden for some reason and I don't have anything to worry about?
 

graysky

Distinguished
Jan 22, 2006
546
0
18,980
Sounds like a hardware monitoring error since multi programs that access the same "probe" gave those numbers. What to do about it? RMA the MB or at least call the manufacture and tell them what's going on with it.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
hmmm, didn't think of that. Yes I was running more than one at a time. I will test this with only one running.

Thanks for the suggestion.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Thanks again everybody, I am going to test this today and see if that is the problem.


I also got a reply back from evga customer support about this issue and they said that all "temperature sensing concerning the processor is dealt with specifically through sensors built in the CPU. Try checking the Intel website for a newer version of driver for the processor as newer version will normally have updated temperature sensing software."


Where would I look for a newer version of a processor driver. I looked on the intel website and couldn't find anything.
 

little_scrapper

Distinguished
Aug 2, 2006
621
0
18,980
Try checking the Intel website for a newer version of driver for the processor as newer version will normally have updated temperature sensing software


Could you actually FEEL them blowing you off?? :)

Cause you dont install drivers for the CPU... thats just stupid.
 

darkwire

Distinguished
Apr 27, 2007
58
0
18,630
Do you run more than one monitoring program at the same time? Doing so can cause conflicts.

I ran Orthos for almost 4 hours today with no errors or weird temp spikes but only used Core Temp, Speedfan, and CPU-z. My temps stayed good the entire time at 3.0ghz. I will try again tomorrow with 3.2 and see how that works.

Maybe I was running to many monitoring programs before and it was causing a conflict like you said.