CPU High Temp

[idata]

Hi all. Just completed a new build with P8H67-M EVO and Core i7 2600K using stock Intel Heatsink/Fan included with processor. Other details probably don't matter, but it is in an Antec Mini P180 case with a Seasonic X-560 (Gold) PS, has 8GB G.Skill Rip Jaws RAM (2x4GB), a Seagate harddrive (SATA) and ASUS DVD Writer (SATA).

After carefully completing the build, I turned on power for the first time, and it powered right up and gave me the BIOS screen. In the BIOS, the CPU temperature shows as 63C-64C at idle! In my opinion this is much too high! I've never seen idle temps this high in any previous builds.

Can anyone suggest any reasons and/or solutions?

[idata]

Hello,

Is a good practice to keep an eye on system temperatures, it will prevent computer problem. Now Those readings are normal for a i7-2600K as the threshold Temperature for it is 72.6°C even though we can see 2 or 3 more °C degrees beyond that specifications when running heavy loads (like gamming and video rendering).

Best regards.

[idata]

Hi Ricardo,

Thanks so much for the reply. I will install Windows and see how the temperatures are then.

[RGiff]

IN your ASUS BIOS you can turn up the CPU fan and bring down those temps on the CPU.

[idata · Accepted solution]

The new ASUS mother boards for Sandy Bridge CPUs are notorious for having rather high CPU temperatures when you are in the UEFI (I have one.) There is a FAQ post in the ASUS forums about this, which states the CPU temperature will be up to 20C higher than normal when in the UEFI. Why this happens is not entirely clear, but it seems the new UEFI "BIOS" puts a high load on the CPU. When in the UEFI, none of the CPU power saving options such as EIST or C-States are active, so the CPU is running at it's standard frequency constantly, and the CPU voltage stays at about 1.2V. Next, the CPU and chassis fan speed control software is not active when in the UEFI, and the fans run at or near their full speed. Given that, it is even more strange that the CPU temperature is that high, but it is. I wonder if that reading is actually correct.

So, the CPU temp you see in the UEFI is normal for your board given the information I have seen and my personal experience with a board similar to yours. I don't use the stock Intel CPU cooler, and my temps are lower than yours. Everything is probably fine, just check the temps after booting into Windows. If you use the ASUS temperature monitoring program, do not run any other programs like that (Real Temp, etc) at the same time, as the ASUS program gets bad data when that is done, and you'll see overheating warnings or very strange temperature readings that are simply wrong.

[idata]

Flashed BIOS to 1850 (the latest for this board at this time), installed Windows 7 Pro 64-bit and all drivers. Haven't installed any monitoring software yet, as I've been struggling with Windows 7 networking, trying to get it to play nicely with XP machines.

Prefer to use the same software across all machines, and have been using SpeedFan on all the other machines for monitoring, but someone elsewhere mentioned CPU-Z, and parsec mentioned Real Temp (which I haven't heard of). Any thoughts on which may be the better choice, and why? [At this point, I haven't even checked to see if any or all of them will run 64-bit and handle this particular mainboard.]

[idata]

It really doesn't make much difference which monitoring program you use, mostly which UI you prefer. They all run on 64bit platforms, or have a version for 64 bit machines, that is nothing new and not an issue. Personally, I like HWiNFO, which works great with my ASUS P67 board. I mentioned RealTemp since it is popular but I don't use it personally.

If your mother board (ASUS?) includes AI Suite II as a monitor, it was pretty good but after a UEFI update, they ruined it. They changed the sensors used for CPU temperature from the on-CPU ones, to apparently ones mounted on or near the CPU socket. I noticed this when after a UEFI update, my CPU temperature at idle was dipping below ambient temperature. Yes, below room temperature, consistently and repeatedly. This was on a high performance CPU cooler, but it is impossible to cool a CPU below ambient with an air or liquid cooler (no I don't use liquid nitrogen.) Why they did this is beyond me, possibly to stop the questions and complaints about the high CPU temps in the UEFI/BIOS. I contacted them about this but was ignored, so I simply don't use the temp monitoring feature in AI Suite II.

This does not affect other temperature monitoring programs, so use what you prefer. I think I recall that I thought RealTemp's temperatures were a few degrees lower than other monitoring programs, although that could have been SpeedFan, and I may have been mistaken. Some users prefer seeing the lower temps, but I prefer reality. This does demonstrate the reality of these programs, they can be adjusted to display anything the programmer (managment) desires. The vast majority of these programs are honest IMO, but there can be exceptions.

[idata]

Thanks for the info, parsec. I will try a few (coretemp, hwmonitor, realtemp, speedfan, maybe others if I come across them) and see which I like best. Will report temps.

[idata]

" If your mother board (ASUS?) includes AI Suite II as a monitor, it was pretty good but after a UEFI update, they ruined it. They changed the sensors used for CPU temperature from the on-CPU ones, to apparently ones mounted on or near the CPU socket. I noticed this when after a UEFI update, my CPU temperature at idle was dipping below ambient temperature. Yes, below room temperature, consistently and repeatedly. This was on a high performance CPU cooler, but it is impossible to cool a CPU below ambient with an air or liquid cooler (no I don't use liquid nitrogen.) Why they did this is beyond me, possibly to stop the questions and complaints about the high CPU temps in the UEFI/BIOS. I contacted them about this but was ignored, so I simply don't use the temp monitoring feature in AI Suite II."

How does your bservation (quoted above) compare with Speedfan's result?

[idata]

As I wrote in that post (in the following paragraph), other monitoring programs are unaffected. As long as the CPU sensors (registers) are read, the temps are fine. Whether or not the change was done in the UEFI or BIOS of other ASUS boards, I don't know.

[idata]

Okay, downloaded and tested CoreTemp, RealTemp, and HWMonitor (one at a time) on a current box (E5200 on an Asus P5Q SE board) alongside the existing SpeedFan. All three were nearly identical for core temps (not surprising, as they all read the same data), and all were about five degrees less than SpeedFan. The installed Asus utility (PC Probe II) doesn't show core temps, but shows die temp and mainboard temp, and those two were identical to SpeedFan. Not sure why SpeedFan was higher on the core temps than the others. Ooops... just realized that HWMonitor is not the same as the HWInfo which you mentioned, so I will give that one a look also.

Having tested those three, I then ran them on the new i7 build, one at a time. What I like about these three is that you can run them without having to install them. Sure enough, they showed core temps around 35C - 38C (ambient 29C), which looks much better. I then ran RealTemp and Prime95 together, and RealTemp's report shows core temps reaching a high of 76C, 79C, 79C and 77C, and finally cooling down to 36C (7C above ambient). Whew!

So you guys were completely correct (as was the article) about the BIOS temps being high, but an O/S able to keep it lower when not at high load. Thanks for the help.

But... I'm not sure what HWMonitor is indicating by the three temps (separate from the four core temps) of SYSTIN, CPUTIN, and AUXTIN, which it shows as 38C, 94C, and 36C respectively, yesterday. The i7 build has nothing installed on it but W7P64 and drivers, so there shouldn't be any conflict. Today HWMonitor shows SYSTIN at 35C, and CPUTIN at 93C (ambient 27C).

On the E5200 build, which does have SpeedFan installed and running, HWMonitor shows SYSTIN at 49C, which is what SpeedFan shows Mainboard (my label) as, and shows CPUTIN at 41C, which is what SpeedFan shows CPU DIE (my label) as. So those seem to make sense. The CPU DIE is showing a few degrees cooler than the cores, and the mainboard a few degrees warmer than the cores. (What does the mainboard sensor measure? Northbridge? Southbridge?)

It is the i7 build numbers in HWMonitor which don't seem to make sense. When I had this box open, immediately after powering down I felt the various components. The CPU HSF did not feel overly warm, nor did the northbridge heatsink or the power heatsinks, but the southbridge heatsink felt very hot.

It is almost as if SYSTIN and CPUTIN are reversed on the i7 box--SYSTIN showing 35C would make sense for the DIE (the same as the cores), while CPUTIN at 93C I'm not sure about, but could be that hot southbridge?

What do you think? Should I be worried about that 93/94C reading (and what is it reading?)?

[idata]

That is a good question, what CPUTIN, SYSTIN, and AUXTIN, really are. I have researched this and never found even a fair answer. The developer of one of the well known utilities whom I asked about these "readings" said he didn't know. Which likely means the answer is not simple, and not worth the effort to explain.

The short answer is they aren't important or applicable to anything. I have seen the AUXTIN really be the "Northbridge" type chip temp, but that is not an absolute. You may also find they match "known" readings, as you did. People have asked about these readings in forums for years, but the only conclusion is that they don't matter. I've ignored them for years and my PCs are fine.

Considering that these hardware monitoring programs are used on many different platforms with all kinds of chipset and CPU combinations, being able to identify them all is impossible, so generic terms are used for the data, as you have seen, and it is up to the user to determine what they actually belong to, if anything.

The main board or mother board sensor is supposed to be a general temperature of the mother board, not of a specific component. That sensor is usually located away from any major heat producing parts, but is not in the same location on every mother board. Some mother board manuals will show the location of that sensor.

In my experience, on i7-900/X58 mother boards, it is the X58 chip (generically speaking, the Northbridge) that runs very warm, while the ICH10 (Southbridge) chip is not as warm, although the latter tend to have smaller heatsinks. Depending on the mother board (and the monitoring chip), HWiNFO will show the X58 and ICH10 temperatures.

[idata]

Hi parsec. I understand there are no longer traditional northbridge/southbridge sets; in my previous post I was referring to the chip in the traditional southbridge location which, on this board, is the H67 chip and could be considered the northbridge. It has a small (in comparison to the other heatsinks) heatsink, and is hot to the touch.

Okay, I've installed Asus AI Suite II, and also SpeedFan on the i7 build. Interestingly, running more than one of these utilities at the same time, in various combinations, doesn't seems to cause any conflicts. So, with AI Suite II, SpeenFan, HWMonitor, and CoreTemp running, I see the following (ambient 24C): AI Suite II CPU: 21C (below ambient, never changes; must be spurious) AI Suite II Motherboard: 31C HWMonitor SYSTIN: 31C HWMonitor CPUTIN: 91C HWMonitor AUXTIN: 30C SpeedFan CPU: 91C SpeedFan System: 31C SpeedFan Aux: 30C The four core temps in HWMonitor and SpeedFan correspond with the CoreTemp readings, roughly 28C to 32C with rapid fluctuations. SpeedFan Aux corresponds with HWMonitor AUXTIN, and this reading does fluctuate slowly and slightly. Since it is in a reasonable range, and does change, I suspect it is a real reading, but don't know what it is reading: Die? Mainboard? Other? SpeedFan CPU and HWMonitor CPUTIN also correspond, and fluctuate more quickly and more widely than Aux. But the reading is so high that it is hard to say whether it is real or spurious. SpeedFan System and HWMonitor SYSTIN correspond with AISuiteII Motherboard, and they fluctuate between the previous two both in frequency and range. This is obviously a real reading, but again, what is it reading: Die? Mainboard? Other? My guess would be that the sensor chip used on the boards provides a number of inputs, and gives them standard names. However, the board designer is free to connect those sensor inputs to whichever sensors he likes, which may not necessarily correspond to the standard names. It looks like both SpeedFan and HWMonitor use the standard names, and therefore match one another, regardless of whether the names accurately correspond to what they are measuring. The core readings seem accurate, and the others (whatever name we give them) seem reasonable, except the CPU/CPUTIN reading, which is much too high. That is the only one I'm concerned about, and don't even know what it is reading, though it does seem to be reading something. For now, I will not pay it too much attention. But I would be interested in the experience of others using this board.

[idata]

eTron, Interesting that the H67 chip runs warm, the P67 on my ASUS mother board runs very cool relative to any Northbridge type chip I have seen on other CPU platforms, socket 775 and 1366 to be specific. Both my G45 and X58 chips are normally at 50 - 60C, and that is cool compared to what others have reported. The P67 apparently operates in the mid 30's C, and it's heatsink is barely warm to the touch, and was less than 100F when checked with an infrared thermometer.

All the recent and current Intel chipsets are rated to have a maximum temp of about 100C (yes, Centigrade) so 50 to 70C is not dangerous.

I can't believe your CPU temp is 91C, from SpeedFan. Intel CPUs have an absolute drop-dead maximum temperature (not Tcase) that is stored in a CPU register, maybe that is what this 91C reading is. What you said about your core temps is what I see, rapidly changing temps in that range at idle or light load. Those are trustworthy IMO, the 91C reading is not.

Any slowly changing temp as AUXTIN and SYSTIN are not the CPU. As you said, either the mother board or I would have said the H67, but the temp is to low according to what you experienced, so I am uncertain.

I see ASUS has ruined the CPU temp monitoring in AI Suite II on all it's boards. They did something in a BIOS update that causes the CPU temp to read ~10C lower than it did, and I saw sub-ambient CPU temperatures at idle. That does not affect other monitoring programs used with my ASUS board, so hopefully that is the same for you.

I still can't run multiple monitoring programs on my ASUS board with AI Suite II, it still goes crazy.

Now I remember why I don't like HWMonitor. I can't comment more on what is what temp and reading wise, since I don't have your board. I know the following may seem to just add more fuel on the uncertainty fire, but I suggest you try HWiNFO. I find it has better identification of readings, although not perfect, and it's CPU readings are correct and labeled correctly. On my ASUS board, it shows both the actual and the "ASUS CPU temp", in different areas of it's display.

IMO HWiNFO will likely help you sort out your readings between all the monitors you have tried. Did you try posting in the ASUS forum for your board, they have separate forums for most boards. You might get some insight from a user in that forum.

It's also possible that you have a bad sensor on the board or CPU, the sensor processing chip is in error, or the CPUTIN reading is not being processed/calculated correctly. For example, on my Core 2 Quad Q8300 PC, monitors show the CPU temperature to be 163C, or 325F, while the cores are in the mid 30's C.

[idata]

Hi parsec, first a correction: previously I wrote that the CPU reading in AISuiteII didn't change. In fact it does change, but very slowly over a long period of time, so I didn't notice any change until the box had been running for a long time. In fact, it seems to change about as gradully as the ambient. But the reading is in fact below ambient, which I'll come back to below.

I double-checked the location of the H67 chip, and it is in fact the one which felt very hot to the touch. I also recall, during extensive research for the components in this new build, reading one review of the P8H67-M EVO board which mentioned this chip running very hot--though I don't remember where, or whether perhaps it may even have been a user review on the egg, or some such (though I think it was a review on a hardware site).

I flashed the BIOS to the latest before installing Windows or any monitoring software, so I can't comment on whether that caused a difference in the way AISuiteII reads on this board. However, in the BIOS, the CPU reading seemed to be about eight to ten degrees lower than with the former BIOS. So they could have changed the way it was reported, or they could have made other changes which caused a lighter load on the CPU when in the BIOS.

But it's interesting that you report the BIOS change as causing AISuiteII to read ten degrees lower, since AISuiteII CPU on my i7 box (below ambient) is about ten degress less than the cores, and also about 10 ~ 12 degrees below the motherboard.

Now here's a question: what does the 'CPU' reading represent? Is it the CPU die temperature? And how is it read? With a thermistor or other thermal sensor?

Okay, though I hadn't reported it, I had already tried HWiNFO on the E5200 box. I had trouble getting it to run. I'm a die-hard command-line type of guy, and use 4NT (formely 4DOS, and more recently known as TCC, a component of Take Command from JPSoftware) extensively. I unzipped the contents into a directory, and when executing HWiNFO32.exe from the command line while in that directory, it would give me errors to the effect that this file or that file could not be found in the current directory. This persisted even after I place that directory on the path. However, I did get it to run using Windows Explorer. I don't know what that did differently.

So on the E5200 box (ambient right now is 24C), HWiNFO has a section called CPU Digital Thermal Sensor, which shows the core temps; they match SpeedFan and the other utilites at ~44C. Then it has a section called Winbond/Nuvoton W83667HG, which shows Motherboard, CPU, and Auxiliary as 46C, 37C, and 13C, respectively; these also match SpeedFan, and as I reported earlier, these SpeedFan readings also match HWMonitor SYSTIN, CPUTIN, and AUXTIN respectively. The difference is that the Aux reading is below ambient on the E5200 box, though not on the i7 box. Even so, the Aux reading on the E5200 box does fluctuate very slightly. But then, HWiNFO has another reading called SYSINT3, which is reading 3C! Other than this last reading, HWiNFO matches the others.

Now on to the i7 box, where I now ran HWiNFO (64-bit). First, the CPU Digital Thermal Sensor section, where the core readings correspond to the others at 30C. Next, there is a new section which wasn't in the E5200 (32-bit) version, called Intel CPU, which has the following temperature readings: CPU Package, CPU IA Cores, and CPU GT Cores, all of which are reading 30C with rapid fluctuations. The CPU GT Cores reading always seems to be equal to, or .5 - 1C below the other two. If I had to guess, I would say CPU Package represented the die, CPU IA Cores represented the four cores, and CPU GT Cores represented the iGPU.

Then comes the Nuvoton NCT6776F section, which shows Motherboard, Auxiliary and CPU readings of 32C, 29C, and 20C, respectively. Motherboard and Aux exactly match SpeedFan. But, where SpeedFan is showing a crazy 93C for CPU, HWiNFO is showing a crazy (because it is below ambient) 20C for CPU, which exactly matches the AISuiteII CPU reading (remember that on the E5200, HWiNFO's CPU reading matched the others and seemed to be a good reading). If we speculate that Asus has done something in the BIOS to make this sensor read ~10C lower than actual, that would make the 20C reading become 30C, the same as the cores and perhaps a correct reading for the die?

Now, still in this section of HWiNFO, inbetween the Auxiliary and CPU readings, there are these three readings: SYSINT1, SYSINT2 and SYSINT3. All three read identically (32.5C right now), and seem to be increasing very slowly. I have no idea what they represent. Hope all this additional data may be helpful to you in forming some opinions about what is going on.

One final speculation. The three CPU readings in the Intel CPU section of HWiNFO64 mentioned above, which don't appear on the E5200 box, hover around 30C ~ 31C, and if all added together, approximate the 90C ~ 93C CPU reading of SpeedFan (and HWMonitor). Is it possible that these utilities are in some way summing those three readings?

[idata]

I don't have the reasons but this could be a solution (http://www.theregister.co.uk/2009/05/11/koolance/ http://www.theregister.co.uk/2009/05/11/koolance/).

[RGiff]

Boyett , you are realy tring to put these people in the deep freeze ?? Or you just having fun ??

[idata]

Not really. I think it is a better solution to their problem.

[idata]

Oh yes, I agree. I use one of those liquid nitrogen cups, but when a mother board is mounted in a standard PC case, I find that the LN spills out of the cup. That does help chill the video card, so not a total loss. Where do you guys get your liquid nitrogen from?

[idata]

parsec wrote:

Oh yes, I agree. I use one of those liquid nitrogen cups, but when a mother board is mounted in a standard PC case, I find that the LN spills out of the cup. That does help chill the video card, so not a total loss. Where do you guys get your liquid nitrogen from?

really?

[idata]

really?

Well, just as really as your other statement:

Not really. I think it is a better solution to their problem.