P4 3.06GHz CPU temp reading

[arch113]

Right now, as I type this, MBM (confirmed in the BIOS too) is reading the CPU temp as (hold on to your shorts) 174F, jumping every now and then to 186F. I know this has to be wrong, I have a temp probe wedge in the fins of the Intel default heatsink, and its reading 81F. The case temp seems to be reading about right (78F). What is the problem?

Motherboard= Gigabyte 8INXP
CPU= P4 3.06GHz (with Hyperthreading)
Heatsick= Intel retail heatsink/fan with AS2 applied.

 

[Bon3thugz43v3r]

Well, try taking out your cpu, then putting it back in, and reseating the heatsink. Your on-die is never off by more than like 10 degrees and I would trust the on-die more than anything else wedged under the heatsink.

 

[arch113]

I have done this once, and it made the temps worse (to what its currently setting at). Before I took the CPU out, cleaned off the AS2, reapplied it and re-fastened the heatsink, I was getting 130-160F+.

Doesn't the CPU have thermal protection? What temp will it shut down?

 

[FastFSB]

WOW THAT HOT!!! The pentium 4 has an on-die thermal diode that is very accurate. I would trust its readings. check to make sure the plastic cage is not dammaged and that the part attaced to the mobo is down completly. i had the cage on the mobo a little loose and my temps went up 20c.

 

[arch113]

I have completly removed the motehrboard, cpu and heatsink, reseated everything, and I still get temps of 80c... When I fisrt turn the system on, and go into the bios. I check the PC Heath screen to see what its temp is, and it reads anywhere from 1c to 116c, I think there is something else wrong here. When I go load MBM it jumps from 80-87 in seconds. I think there is something wrong, but don't know if its the motherboard or CPU?

 

[FastFSB]

if its jumping around like that i have a feeling that the problem is not just heat, but maybe a bad reading. Maybe the thermal diode in the cpu is defective. maybe the board is causing the problem. do you have any other boards you can try the cpu in? Or do you have any other cpu's to try in the board? If not, you might want to concider RMAing the board and/or cpu. pentium 4's throttle themselves down if the temp is too high. does the system feel sluggish in windows?

 

[arch113]

This is a fresh load, so I don't have much on here, web browsing seems ok

How much will the CPU throttle back? Would Gigabyte's EasyTune4 show that the MHz has changed?

 

[FastFSB]

the throtteling varies on temp... intel's site gives more tech details but basicly it just skips a bunch of clock cycles. easy tune wont report it. try running sandra and see if your cpu score goes down drasticly. it is also possible (but probably not that likely) that the throtteling mechanism is not working. i doubt this is the case but i want to get all the possiblites out there for you to concider. try running prime95 and see if there are any errors. if your cpu was really that hot i have a feeling it would produce errors pretty quickly

 

[modenaf1]

the cpu probably isnt that hot because even without a heatsink the p4 should throttle down far enough to keep it away from those high temps.

 

[arch113]

I sent an e-mail to Intel and Gigabyte tech support.

I will see who will blame the other first......

 

[I.M.O.G.]

A common problem with mbm5 is that people do no have it configured correctly.

Do your temp readings respond to load changes on the cpu?

I do not know what pentium heat ratings are but the readings you are getting are definetly in the danger range for amd's and I think pentiums safe range is lower than amd.

178F=81.11111111111111
186F=85.55555555555556

Your chip would probably be dead already, or could be very close to it. The absolute max for amd's are 95C, and that equates to instant death I believe.

How was your computer behaving in various applications? Normally? Did you get any errors or instability?

I am betting you are just getting bad readings from software misconfiguration. MBM5 comes with a readme that explains how to adjust the settings correctly. I never looked at the directions but just played around until it read my sensors reasonably and responded to cpu load like I knew it should.

This isn't what's causing the problem, but it will make a difference in the quality of temps you finally get... did you apply the as2 according to the directions on the arctic silver website?

 

[arch113]

The reading from MBM is the same as the reading from Gigabytes Easy Tune 4 application, which is about the same as it reads in the BIOS. So far there has been no discoloration on the CPU or from AS2. The system has locked up on me one yesterday while installing the 3rd cd for Microsoft Flight Simulator 2002 Pro. I did play BF1942 for 15-20 minutes with no problem....

 

[FastFSB]

have you tried prime95? also, go into bios and lower your fsb to the absolut min allows (100mhz). see if your temps go down. it really seems like it is a problem with the temp sensor or the mobo's handeling of the on-die temp probe. i have a feeling that your cpu would produce errors left and right if those temps were accurate, which is why i think running prime95 in stress test mode is such a good idea.

 

[arch113]

I didn't lower the fsb, but I did run prime95 for 1hr 45 minutes, the only thing that crashed was the Easy Tune 4 application. MBM cpu temp went from 77c to 89c The temp probe I have on the heatsink went to 109F. I still haven't heard from Intel or Gigabyte.

 

[FastFSB]

ok. it seems like your cpu would produce a lot of errors or throttle down if the temps were really that high, although 1hr45min is not that long... run it agin overnight just to be sure. i've had the gigabyte easytune program crash on me before too, so i wouldnt worry about that. the temp probe on the hs seems about right for prime95 on a 3.06. try calling them on the phone. intel and gigabyte seem to respond better that way, at least from my experience. i emailed gigabyte and never got a repsonse, but i called and solved the problem in about 10 minutes. is there any way you can get another p4 to test in your board?

 

[arch113]

I have been running Folding@Home for 6+hours so far, and even played BF1942 online at the same time, and no problems so far. The temp didn't even raise much. I wish I had another P4 CPU to test, but I don't, I have mostly AMD's around here.....

 

[FastFSB]

i would really try the prime95 self test. if the cpu is producing erros in folding, they may go unreported. if it produced errors in games, you may not notice. try prime95 over night or for at least 12 hours. i still believe that the problem is with the temp reporting, not the temperature being too high.

 

[Samoyed]

Rather than waste time and lose your sanity I advise several steps:

1) Touch the bottom edge of the heat sink. If it is really 80C you will burn your finger if u hold it there for more than a few seconds. If u dont then your temp reading is wrong.

2) Does the comp run stable? How about if u overclock it a bit say 10 mhz FSB. Still stable? If so who give a rats *** how hot it "reads"? It's probaby reading wrong.

3) Intel CPUs will shut down before they are destroyed..have no fear about that.

4) BY FAR the best way to isolate your problem is to get another cpu that fits that socket...any CPU and see if the temps are normal. If so its probaby your CPU. Next try your CPU in someone elses board that has Temp reading ability. Normal temps mean your MB or bios is reading wrong. High temps indicate the CPU is at fault.

 

[FastFSB]

Rather than waste time and lose your sanity I advise several steps:

1) Touch the bottom edge of the heat sink. If it is really 80C you will burn your finger if u hold it there for more than a few seconds. If u dont then your temp reading is wrong.

2) Does the comp run stable? How about if u overclock it a bit say 10 mhz FSB. Still stable? If so who give a rats *** how hot it "reads"? It's probaby reading wrong.

3) Intel CPUs will shut down before they are destroyed..have no fear about that.

4) BY FAR the best way to isolate your problem is to get another cpu that fits that socket...any CPU and see if the temps are normal. If so its probaby your CPU. Next try your CPU in someone elses board that has Temp reading ability. Normal temps mean your MB or bios is reading wrong. High temps indicate the CPU is at fault.

ok, lets assume that there is poor contact between the cpu and the hs. even if the cpu was at 100c (i know it would burn up but for the sake of agrument stay with me), the hs could be COLD to the touch. you cant just measure the hs temp without knowing that there is good heat transfer. im not saying that there is a styrofoam block inbetween the hs and cpu, but the clamp pressure could be poor, and the throttleing mechanism could be defective. althought this is not the most likely case, there is no harm in throwing out the possiblilty, but there could be harm in ignoring it.

 

[FastFSB]

sorry, hit return accidently. anyway....

secondly, running prime95 is an excellent way to test stability. just because you can "get to windows" doesnt mean your system is stable. i can get to windows at 3.7 but its not producing error free results. small problems such as a miscalculation here or there are indicitive of bigger problems. assuming that the temp sensor is "probably reading wrong" is a good way to fry an expensive cpu. i think that running prime95 or a similar program overnight is worth a $600 cpu, dont you?

as i stated above, it is possible but not likely that the throtteling mechanism is defective. p4's dont actully shut down, but they skip a lot of clock cycles to prevent overheating, even if you take the hs off during a quake 3 demo. tomshardware.com has a video of this.

i agree that another cpu would be a good way to test it, which is why i posted that earlier.