Are these temperatures too high (CPU)?

Factual Questions
1

I’ve been having an ongoing problem with computer crashes that I’ve described here. Since I last posted on the subject, I’ve found that disabling my onboard sound did not help, nor did buying a new graphics card (Radeon 4850).

One of the suggestions in the other thread was that my CPU is bad, so last night I decided to try stress testing my system with Prime95; I set it running and went to bed. When I awoke this morning I was not at all surprised to find my system had crashed/rebooted at some point during the night… I was surprised, however, to find that it had happened after only about 25 minutes.

“If at first you don’t succeed…” I fired up Prime95 again, and ran SpeedFan alongside it to see what was going on, and took a few screenshots. After a few minutes; fifteen minutes later; and a half-hour after that. At that point I stopped the tests - I know, I know, you’re supposed to let it go for several hours, but I have other things to do. I’ll try running it overnight again. I took one final shot a few minutes later, with the system at idle.

I mentioned in the other thread that I’m not sure how much SpeedFan can be trusted. For instance, I assume that my actual CPU temp is Temp3 or Temp4, not the -50C reported under CPU. And it says Fans 2 and 3 are at 0 RPMs… what do these mean? There are four fans inside my case, if you count the CPU fan and the GPU fan; the two case fans are most definitely running. I can simply glance down at the grates on my box to affirm this.

Anyway, as I stated in the other thread, I thought I had ruled out overheating as a cause for my problems, but after looking at these SpeedFan readings I’m not so sure. I’ve been Googling a bit on the issue, and I’ve come across all sorts of people, with systems similar to mine, reporting (and/or showing screen shots of) much lower core0 and core1 temperatures. Like, in the 30s, where mine average about 55C at idle. I’ve also read from some people that 55 is nothing to worry about, others say it’s too high.

Oh, and I’ve been checking the temperature in my BIOS quite a bit lately - it always reports my CPU temp at about 45C, with a margin of 35-40.

Why am I getting a difference of 10 degrees between my CPU and the cores? Aren’t they the same thing, or, aren’t the cores components of the CPU? They’re in the same place, how can the temperature vary so wildly?

And most importantly: are these temperatures too hot? Could this be why my system has been crashing?

Footnote: someone suggested, in the other thread, that my power supply could be the culprit; if SpeedFan is to be trusted, it appears my PSU is ok…?

2

Don’t know enough about hardware to comment on the correct functioning of the internal fans or the other inner working of your computer, but I do know that computers and heat do not mix.

What is the ambient temperature of the room it’s in?

Computers will do some really freaky and funky things starting at about 80F

3

I’m not familiar with Speedfan, but your third screenshot shows multiple temps in the 80’s celcius. This is not good.

I would open the case and ensure that the CPU fan is running. If it is, then I might suspect that the heatsink is not seated properly. Removing and reinstalling the heatsink might take care of it.

As for fans showing 0 RPM, only fans that are attached to the 3-wire connectors on your motherboard will report their speeds. Some of your fans might be attached to hard disk power connectors from the power supply.

4

All of the fans are functional; the heatsink is seated properly; the inside of the case is cleaned on a regular basis; etc. Any other preventative measures I’ve taken are described in the other thread.

As for the screen shots, bear in mind the first three show the CPU being pushed to its absolute limit. This should never happen under ordinary circumstances. It’s the last screen shot, the idle one, that concerns me the most.

It’s currently 68F in my apartment, and it rarely gets much warmer than this in here. On the rare occasions it hits 90 or 95 outside, it might get up to 75 or 78 inside - nothing that should be contributing to my problem.

5

The labels in Speedfan can’t always be trusted, but looking at your screenshots it would be reasonable to conclude that you have a case temperature in the 40s and your CPU is rising over 80 under load (no doubt it continued to climb in your previous stress test, until the system cut the power to protect itself).

The high case temperature suggests that it’s not just a problem with your CPU heatsink or fan. I would try running the computer without the case to see if the temperatures improve. If so, the problem is airflow. Of course if you have an extremely high ambient temperature in the room then that could be the reason for your problems.

Edit: just saw your post - your case temperature shouldn’t be that high with those ambient temperatures. Definitely try removing the case, and make sure all of the fans are running while you’re at it (but don’t touch anything while it’s running!).

6

Did you configure speedfan by downloading the correct profile for your motherboard? Given the generic labeling, and the totally whacked temperature being reported for the CPU, I’m guessing maybe not. A good profile for your board should have (mostly) informative labels and reasonable thresholds set. That said, a core temperature of 80-something Celsius sounds like you’re frying the poor thing. Offhand, I don’t know the range for your processor (I’m running an AMD), but I would be surprised if that high was OK.

7

Intel’s max safe temp is is like 70 degrees, so running 85+ at max load means your case can use an extra fan, but its not really a big deal. If the computer isnt locking up and crashing when prime95 is running then you probably dont have a heat issue.

As far as accuracy, install coretemp and compare. You should get the same readings from both.

Thats normal. One core may be busier than the other.

As far as troubelshooting goes, dont uses memtest86, use the microsoft memory diagnostic instead. Also, have you checked for your hard drive yet? Do a chkdsk. Check your smart values, speedfan can do that now.

That said, I still dont understand what kind of crashes you are having. Is the computer just locking up? Rebooting? Windows can be set to do a memory dump at crash time and you can analyze that dump with Microsoft’s tools. Ive done this in XP many times, but never in Vista. It can usually point to what driver or device is crapping out on you.

8

Just wanted to say that with my last comp, I apparently didn’t put the thermal paste on right or possibly it just lost grip, and the heatsink detached just enough from the CPU to give me problems. Since thermal paste is only a few bucks, you might want to try taking your heatsink off, cleaning all thermal paste from sink and cpu, then reapply the past and put the heatsink back on.

Not sure if you already checked the memory sticks, but you might want to run with just one for a week, then take it out and put the other one in by itself for another week. Long shot, but it’s something you can do to rule out a memory issue, and won’t require spending more money.

Good luck, issues like this can be a pain to diagnose.

9

btw kb on analyzing memory dumps here:

10

I’m actually running it with the case open right now - after I made my last post I powered off and reseated the heatsink, just to be sure. All of the fans are running. The temperatures are about 5C lower, we’ll see what happens when I seal it back up.

Yeah, I’ve been using that too. It actually reports the cores as being 5C cooler than Speedfan, and they stay in step.

Heh… everything I’ve read over the last year says ignore Microsoft and use MemTest.

It reboots, without warning. I have Windows set to give me a BSOD when it crashes, but that never happens.

In the past few days my go-to program to actually get my system to crash is NWN1’s toolset (the game itself, however, does not cause any problems). I’ve had the toolset running for the last 20 minutes or so (since reseating the heatsink), with all the graphics/sound options turned on to increase the stress, and I’ve got SpeedFan and Coretemp running. So far, so good…

I can’t believe that’s the problem, though. The heatsink wasn’t loose or anything before, it seems exactly the same now. I suppose I should shut down and seal the box back up, and go from there.

11

In CoreTemp, The difference between Tj. Max and the core temperatures is the thermal margin, correct? So, for instance, right now my core temps are both at 58C, and Tj. Max is 100C. So this means my CPU isn’t supposed to overheat until the cores get to 100C? That seems hard to believe.

Strangely, I think I’ve figured out how to deliberately crash my machine. Using the NWN1 Toolset, I’ve noticed a few of my crashes have happened when I selected Options from the Tools menu. So I tried firing up the toolset, using the mouse wheel to continuously zoom in and out, then after a couple of minutes of this trying to bring up Options… voila! Works like a charm!

I kept an eye on SpeedFan and CoreTemp when I did this. The crash happened when the core temperatures (according to CoreTemp) got to 80 and 79.

I’m gonna try to duplicate this a couple more times.

12

Well, after doing some more Googling, I determined that a big chunk 'o my problem is that I’ve been using the fan & heatsink that came with the CPU. I went out and bought a Big Ass ThermalTake fan, installed it correctly (i.e., new thermal paste), and finally got my system running again a little while ago…

So far the core temps have been hovering around 45C at idle. I ran Prime95 for a few minutes, and the temps fluctuated between 62 and 65, didn’t go any higher (unlike this afternoon, when they immediately shot up into the 90s). The BIOS reported the temp when I first powered up at 28, SpeedFan is currently saying 38. Much, much better.

However, I tried my trick with the NWN toolset… the core temps never went above 58 while I was zooming in and out, but when I tried to bring up Options my computer still fucking rebooted. But at this point I’m gonna assume there’s a problem with the toolset, and it crashing my system in the same way is just a coincidence. I’ll just have to wait and see if it crashes at another time.

In conclusion, it remains to be seen if my problem has been solved. :frowning:

13

In the 90s?! Celcius? Holy moly. I’m going to say that this extreme overheating has probably damaged something, and it’s that which is causing your problem now.

14

If you go above 70 deg C, your CPU is going to die in a very short amount of time. 70 deg C is not the temperature you want to stay under, though.

A rough rule of thumb is that every 10 deg C above room temperature cuts the life of the CPU in half. This quick and dirty rule is a bit inaccurate, especially at either end of the scale. Above 70 deg C the CPU dies faster than this rule predicts, and below 45 deg C or so there’s no real harm to the CPU even though it is above room temp.

45 deg C is really the number you want to aim for. Most CPUs these days run a bit hot, and most people replace their computers after only a few years, so if your CPU runs at about 55 deg C or so it is no big deal.

60 is a bit hot, but you shouldn’t have computer crashes that result. The OP’s CPU has been overstressed before though, so it may already have suffered some damage.

15

Sigh… Is there any way to know for sure?

I tried running Prime95 last night before going to bed, with the intent of letting it go overnight. Twice. Both times my computer crashed within a minute. Is that my indication that my CPU is toast?

16

You didnt say whether you’d removed the Audigy, or if you’d tried one stick of ram at a time, and swap.

My bet is the Audigy, they’re fairly legendary for causing problems with hard drive controllers.

Otara

17

Oh yeah, I did try removing the Audigy. No luck. I haven’t tried the sticks yet, but I did run both Microsoft’s memory diagnostic and Memtest86 (again), neither report any problems. Thanks for the reminder, I’ll try removing the sticks.

I also ran chkdsk on my hard drive today (the full check, taking several hours), and it came up with a few bad sectors. I haven’t had time yet since to try crashing my computer…

18

Sod.

Well you’re into that horrible situation where you just have to keep swapping out till you find which component it is. Could be CPU, MB, power supply, videocard, ram, unlikely to be OS (ie reinstall) but could be. Good that you eliminated heat.

Loose/broken solder connection that separates under heat, who knows

Bad sectors on HD isnt too surprising if you’re getting hard reboots but fingers crossed for you. This is your worst scenario, where you dont have a previously known good situation to work from.

Otara

19

Well, my trick to deliberately crash my computer worked as usual…

I’m gonna try removing the ram sticks tomorrow, but if that isn’t it, it’s either my CPU, mobo or Vista - everything else has been eliminated. That is, I think the power supply has been eliminated - the voltages reported in SpeedFan seem to indicate it’s OK, and my BIOS reports the same numbers.

20

The other day I downloaded speedfan and took the first serious look at my temps in a while. While most of the system (including the CPU) was running at acceptable temps, I noticed that my Geforce 8800 had a temp of 80C! This freaked me out because I started imagining cooking breakfast on my video card, but then I found out that there was a program called Riva which would truly adjust my GPU’S fan. Turns out it was only running at 37% speed. I boosted it to 100 and got used to the extra noise in 5 minutes and watched it drop down to a more healthy 50C.

You might want to check that your fans are running at 100% too.