New coretemp show Incorrect temp(I think)

[Doodies]

I have an e2180 M0 stepping cpu and it is now displaying 43-45c idle temp and 62c+ under load (orthos). Even when I clocked it down to 1.6ghz and at 1.2 vcore it was displaying 38-39c idle. These temps seem far to high for this cpu with a Better than stock HSF (ocz vendetta). If I look at review for this HSF they show similar cpus over clocked and they never get this hot except for Quadcores that have been over clocked.

Here is a pic of it idling at 1.6ghz



I know in this release that the tjmax has been raised for these cpus but the temps don't seem right.

[The Coolest]

I guess the Tj. Max wasn't raised. In the next version it'll be back to normal.

[GigaByte]

I have an e2180 M0 stepping cpu and it is now displaying 43-45c idle temp and 62c+ under load (orthos). Even when I clocked it down to 1.6ghz and at 1.2 vcore it was displaying 38-39c idle. These temps seem far to high for this cpu with a Better than stock HSF (ocz vendetta). If I look at review for this HSF they show similar cpus over clocked and they never get this hot except for Quadcores that have been over clocked.

Here is a pic of it idling at 1.6ghz



I know in this release that the tjmax has been raised for these cpus but the temps don't seem right.

The new CoreTemp readings for M0 steppings are correct, read this for more info: http://www.overclock.net/intel-cpus/263 ... ronix.html

The Tj max is not 85c for these series of procs, Tj max is the temp the CPU will shutdown at, let your CPU heat up to 85c and you will notice it will NOT shutdown, but will shutdown at 100c because the Tj max is 100c. As to why so many programs read it as 85c is because I believe Intel made a last minute change to M0 (and a few others possibly) procs by raising their Tj max to 100c for whatever reason, and in the guide I posted you can verify yourself that it IS indeed 100c Tj max. CoreTemp shows the CORE temps while your thermal spec is based on TCASE and Tcase is ALWAYS 10c (+/-3c) lower than the core temps. Some processors have a 15c (+/-3c) delta between the cores and the Tcase.

Just read the guide above for complete info, it will make sence once you read it, yes the whole thing. I am going to save a copy of CoreTemp 0.96 incase Tj max is reduced to 85c again which is incorrect...

[DrNinjaman]

I put together a system based on the E2160 pentium dual core M0 stepping processor and was doing some overclocking on it using the previous version of CoreTemp (0.95.4). I did notice the Tjunction Max being reported as 85deg, but thought nothing of it, though I knew this processor has the higher Tjunction Max of 100.

Upon switching to CoreTemp 0.96 the temperatures are being reported significantly higher.

So my question is: which one is right? 10 degrees C is a big difference, and I'd like to know how hot this thing is actually getting.

[GigaByte]

I put together a system based on the E2160 pentium dual core M0 stepping processor and was doing some overclocking on it using the previous version of CoreTemp (0.95.4). I did notice the Tjunction Max being reported as 85deg, but thought nothing of it, though I knew this processor has the higher Tjunction Max of 100.

Upon switching to CoreTemp 0.96 the temperatures are being reported significantly higher.

So my question is: which one is right? 10 degrees C is a big difference, and I'd like to know how hot this thing is actually getting.

CoreTemp 0.96 is correct, M0 steppings are 100c Tj max. Check my thread here with proof: http://www.overclock.net/intel-cpus/277 ... 0c-tj.html

[Doodies]

I put together a system based on the E2160 pentium dual core M0 stepping processor and was doing some overclocking on it using the previous version of CoreTemp (0.95.4). I did notice the Tjunction Max being reported as 85deg, but thought nothing of it, though I knew this processor has the higher Tjunction Max of 100.

Upon switching to CoreTemp 0.96 the temperatures are being reported significantly higher.

So my question is: which one is right? 10 degrees C is a big difference, and I'd like to know how hot this thing is actually getting.

CoreTemp 0.96 is correct, M0 steppings are 100c Tj max. Check my thread here with proof: http://www.overclock.net/intel-cpus/277 ... 0c-tj.html

Okay I think I agree with you, but if you look at the intel website they say the e2180 max temp threshold is 73 degrees which if you lowered the throttling temps you got by 15c would more accurately reflect that at around 80c and why are these cpus running so much hotter than all of the other core2 cpus?

[The Coolest]

It appears that some CPUs have a 100C Tjunction Max, even though the CPU still reports 85C. But most chips do, most likely, have 85C Tjunction Max and from some feedback I've had I decided that the next version will go back to the original system of Tjunction Max detection for the M0 revision.

[Doodies]

I have a question, if your over clocking what do you think is a safe delta between the temp and actual tjunctionmax? I've been using that cause no matter what the tjmax actually is the difference is the same.

Also another thing about the tjmax being 85 instead of 100, There were some processors I believe that were idling below ambient which are the cpus that need the tjmax raised I think one of them was the e6850..

[The Coolest]

I'm aware of the fact that some CPUs do report sub-ambient temps while still reporting the Tjunction Max as 85C. That makes no sense.
That's why the L2 and G0 revisions are hard-coded to 100C Tj.Max no matter what the actual processor reports. I tried to do the same for M0, but from the reports I've been getting it seems that a lot of people are reporting too high temps.

A rule of thumb, not sure how accurate\reliable it is, is to keep the core temps 15-20C Tjunction Max for a reliable stability and not to shorten the CPU lifespan.
I personally ran my chip, full load, at 65-75C for several months in a row and it's still alive and running stably.

[cpmee]

A rule of thumb, not sure how accurate\reliable it is, is to keep the core temps 15-20C Tjunction Max for a reliable stability and not to shorten the CPU lifespan.
I personally ran my chip, full load, at 65-75C for several months in a row and it's still alive and running stably.


I just recently setup a Compaq notebook for a friend that uses the T2310 chip. It uses the exactly the same family and stepping, M0 , that my e2180 has. The temps using CoreTemp 0.95.4 were 77C on Orthos load. I thought that was high until I searched on the web and saw that other notebooks with that chip were even higher. So I believe the M0 stepping is built to take the heat, so the 100C Tj.Max figure has to be correct.

However,

For the record, my e2180 maxed out at around 51C at 2.8ghz on the stock intel heatsink and artic silver ceramique with 0.95.4

With 0.96 though, the temps show around 67C. Based on my old reliable finger touching the heatsink method of determining temps, there is no way 67C is correct. I wouldnt be able to hold my finger on the base of the sink for longer than 20 seconds if it were. But I can, and it only feels mildly warm at load.

Please go back to the old 0.95.4 method, which is much closer to TAT and other program readings. I believe they are closer to correct.

[LuxZg]

A rule of thumb, not sure how accurate\reliable it is, is to keep the core temps 15-20C Tjunction Max for a reliable stability and not to shorten the CPU lifespan.
I personally ran my chip, full load, at 65-75C for several months in a row and it's still alive and running stably.

Is that 15-20C delta Tmax under real-world load, or under OCCT/Prime, or under completely artificial load like Intel TAT?

I have delta Tmax with CoreTemp 0.96 no less than 17C under TAT 100% load, and no more than 28C with OCCT for example..
Real world loads should be more like OCCT's temps, right?

So if your "rule of thumb" is under these loads, than I still have 5C space for overclocking...? Ofcourse, if system stability can be achieved, that's another story..

[The Coolest]

As I said, it's a rule of thumb...
I personally wouldn't like my CPU to go over that threshold at full synthetic load.

[LuxZg]

Ok, thanks for a REAL quick answer. I'll stay at this clocks, it's enough anyway..

But I'm glad I've found this thread, as previously I was trully worried with both TAT and CoreTemp showing 83C with Scythe Ninja on only 3.2GHz/1.38V dual core CPU That's on full TAT load over 20 minutes..

Thanks for explanations throught this thread, it made me feel better

[nosam]

I have an e4500 which is obviously m0 stepping. Coretemp reports about 45-50 idle but... BIOS, Gigabyte EasyTune, Everest, HWMonitor, TAT. all report about 30-35idle. I am still REALLY confused! Seem 5 to one against coretemp but it doesnt mean the majority are right.

Also it seems that if you are pushing your cpu to 90 with a program using Tj Max=100 then thats pusing it to 75 with Tj max=85. So i really dont know how you could logically test it. either way you would be 10 degrees under Tj Max, because thats how the apps seem to me to be working.

[GigaByte]

I don't know how much more proof is needed... but if you really want M0 back as 85c Tj max then so be it... ill be keeping a copy of 0.96 because its accurate..

http://www.hardforum.com/showthread.php ... 806&page=2

http://answers.yahoo.com/question/index ... 804AASApEg

http://whirlpool.net.au/forum-replies-a ... 88124.html

http://www.tomshardware.com/forum/24517 ... ping-e2180

http://www.overclock.net/intel-cpus/277 ... 0c-tj.html

http://www.overclock.net/intel-cpus/265 ... p-0-a.html

http://www.ocforums.com/showthread.php?p=5411537

http://www.xcpus.com/forums/power-suppl ... guide.html

http://www.sk.co.yu/forum/showthread.php?p=433013

http://translate.google.com/translate?h ... f%26sa%3DN

[nosam]

So does this mean my e4500 actually idles with c1e enabled so only 2400mhz at 45 with my room temp at about 20 and an arctic freezer 7 pro, with load (3200mhz, and 1.3625v in bios) 65+. either way if i get to 85 in everest or 100 in coretemp it will be the same temperature so it doesnt really matter which is which aslong as i say a good 20 below tj max for that app. Am i thinking along the right lines here?

[Doodies]

I don't know how much more proof is needed... but if you really want M0 back as 85c Tj max then so be it... ill be keeping a copy of 0.96 because its accurate..

After reading through all of those, it seems to me that about half the people still think the temps aren't correct. Some are saying that these cpus are tjmax 100 but the temps shouldn't have 15c added and others are saying that these cpus actually have a tjmax of 85c. Here is my question to you, if in fact these are the real temperatures, why are these M0 stepping cpu's running so much hotter then any other core2 cpu? Even the Quad cores are running cooler then these cpus.

Look at the pic of my under-clocked and under-volted cpu idling at 38-39c with a aftermarket cooler, how do you explain that?

[GigaByte]

I don't know how much more proof is needed... but if you really want M0 back as 85c Tj max then so be it... ill be keeping a copy of 0.96 because its accurate..

After reading through all of those, it seems to me that about half the people still think the temps aren't correct. Some are saying that these cpus are tjmax 100 but the temps shouldn't have 15c added and others are saying that these cpus actually have a tjmax of 85c. Here is my question to you, if in fact these are the real temperatures, why are these M0 stepping cpu's running so much hotter then any other core2 cpu? Even the Quad cores are running cooler then these cpus.

Look at the pic of my under-clocked and under-volted cpu idling at 38-39c with a aftermarket cooler, how do you explain that?

Try to put this as simply as I can..

1. Tcase is temp measurement in the middle of the IHS, this is what Intel rates their thermal spec on (Tcase max or Tc Max) not the core temps.

2. Tcase is always lower than core temps on every current duo and quad.

3. Tcase is always higher than ambient, even if its 0.1c above it.

Now, depending on which stepping CPU you have your Tcase to core temp delta is 10c or 15c (+/-3c), in our case all M0 steppings have a 10c (+/-3c) delta. If Tj max is 85c for the M0 stepping that will make the core temps +/-2c from the Tcase, which is NOT possible. Having that said some will think Tcase is off and needs a minus offset, this IS possible but the chance of a CPU literially blowing up into a million pieces is more likely than this. So if you give Tcase a -10c offset to keep the 10c delta that would make your Tcase lower than ambient, and theres the problem.

Lower than ambient, NOT possible even under water cooling.

EDIT: To why M0 stepping is so much hotter than other steppings is not very understood by many if anyone, including myself. They do run on less vcore than L2 steppings and generally clock higher (most the time). But they do have a higher thermal spec and I would guess Intel raised the thermal spec due to the ammount of heat M0 CPUs give off. Remember thermal spec is Tcase, not core temps. Tcase is read as "CPU" in most programs and if your Tj max is read correctly or you set an offset +15c for the cores you should see Tcase is always about 10c lower than the cores. Under extreme overclocking and vcore the delta may exceed 16c which is normal.

[Doodies]

I think I understand most of that...

There are two things that are throwing me off. One is when you got your cpu to throttle the tcase was at 86c and the cores were at 97, shouldn't it have throttled at around 73c and probably shut off by 86c. Intel spec says max tcase is 73c. The other thing that is throwing me off is that my heat sink isn't getting that warm, My last cpu p4 Prescott would have my heat sink a lot warmer even at idle than my new cpu is under load.

[GigaByte]

I think I understand most of that...

There are two things that are throwing me off. One is when you got your cpu to throttle the tcase was at 86c and the cores were at 97, shouldn't it have throttled at around 73c and probably shut off by 86c. Intel spec says max tcase is 73c. The other thing that is throwing me off is that my heat sink isn't getting that warm, My last cpu p4 Prescott would have my heat sink a lot warmer even at idle than my new cpu is under load.

Tcase max is indeed 73c, but Tcase is a temperature measurement at the center of the IHS (metal plate that your heatsink sits on). Tcase is not a core or a heat source, the only heat source are the cores. The cores throttle 5c below Tj max and system shutdown occurs once Tj max has been reached. Tcase temp even above spec is not what controls throttling or shutdown. The PROCHOT# signal is what starts/stops throttling and the THERMTRIP# is the signal that will shutdown the system once Tj max is reached. The only signal, PROCHOT# can be forced to stay inactive if the cores reach throttle temp by disabling TM/TM2 in bios. But THERMTRIP# is not accessable or alterable by bios or any software for the sake of the CPU's safty. THERMTRIP# basicly watches the core temps and if they reach Tj max (85c or 100c depending on stepping) it will cut power to the cores, hense system shutdown. Lets say if you are running at 75c Tcase, yes you are over spec and your cores should be at about 85c (+/-3c), but no more than 92c hands down. Your cores have not yet reached throttle temp as that is 95c for M0 but you have already exceeded Tcase max. Thats why the Tcase is the most important temp to watch (cores are important as well) when load testing, you will always have Tcase as your limit before the cores themselfs.

While the cores were at 97c the heatpipes were quite warm on my Zalman CNPS9700 NT, both it and the CPU are lapped using AS5. I have tried my CNPS9700 with my Pentium 4 630 @ 4Ghz 1.52v and the heatpipes didn't even feel slightly warm, but the P4 was on load @ 49c. I didn't think that would be a good test for heat transfer as the cooler is extremely high end for Netburst CPUs since it just drags the heat out so quickly it has no chance to build up. One reason that could result in high M0 stepping temps while some heatsinks don't feel that hot could be poor seating between the die and the IHS. Lapping the IHS alone helps alot as almost every duo and quad is concave.

[nosam]

How did you measure the temperatures? Do measuring tools not display temps depending on a delta from tj max anyway. So, doesnt this mean that whatever the program calls tj max will basically 5 degrees above when the program reports the cpu throttling.

You may well be right, but dont you think idle temps in the range of 45-50 are a little bit excessive for modern processors? Even if i run it at a lower vcore and frequency than an L2 chip it is still hotter. To me thats non-sensical.

Just to make it clearer. If i did what you did with my chip in everest (reports tj max =85) i would see throttling at 80. Would that be proof?

[GigaByte]

How did you measure the temperatures? Do measuring tools not display temps depending on a delta from tj max anyway. So, doesnt this mean that whatever the program calls tj max will basically 5 degrees above when the program reports the cpu throttling.

No, programs like Everest and others that monitor throttling only report throttling once the PROCHOT# signal is active.

You may well be right, but dont you think idle temps in the range of 45-50 are a little bit excessive for modern processors? Even if i run it at a lower vcore and frequency than an L2 chip it is still hotter. To me thats non-sensical.

Modern processors are only going to get hotter as they get faster and able to overclock more, remember the Pentium 133Mhz? Those ran without a heaksink and were barely warm. Sure modern processors will get hotter, but they also take more heat, ex 60.1c Tc max on L2 and 73.3c Tc max on M0.

Just to make it clearer. If i did what you did with my chip in everest (reports tj max =85) i would see throttling at 80. Would that be proof?

Read what I posted above, this is how 100c Tj max prove correct.

Now, depending on which stepping CPU you have your Tcase to core temp delta is 10c or 15c (+/-3c), in our case all M0 steppings have a 10c (+/-3c) delta. If Tj max is 85c for the M0 stepping that will make the core temps +/-2c from the Tcase, which is NOT possible. Having that said some will think Tcase is off and needs a minus offset, this IS possible but the chance of a CPU literially blowing up into a million pieces is more likely than this. So if you give Tcase a -10c offset to keep the 10c delta that would make your Tcase lower than ambient, and theres the problem.

Lower than ambient, NOT possible even under water cooling.

[cpmee]

While the cores were at 97c the heatpipes were quite warm on my Zalman CNPS9700 NT, both it and the CPU are lapped using AS5. I have tried my CNPS9700 with my Pentium 4 630 @ 4Ghz 1.52v and the heatpipes didn't even feel slightly warm, but the P4 was on load @ 49c. I didn't think that would be a good test for heat transfer as the cooler is extremely high end for Netburst CPUs since it just drags the heat out so quickly it has no chance to build up. One reason that could result in high M0 stepping temps while some heatsinks don't feel that hot could be poor seating between the die and the IHS.

That may be true with heatpipe heatsinks due to the cooled fluid draining back, but the stock intel one is all aluminum. The heat would conduct straight out and be felt. At what version .96 is telling me, 44C, it would feel very warm to the finger. Yet, its downright cool. Version .95.4 says 29C, which is still above ambient. Version .95.4 is also very close to intel TAT temps for me, 30C, (e2180 M0 stepping). In my case, version .95.4 temps seem like the correct ones from all the evidence.

There could be poor seating between the die and IHS, but the heat must conduct out. (unless using an insulator for thermal paste )





Lower than ambient temps are very easy to do with watercooling, all you need is either a bong cooler or ice in the resevoir.

[nosam]

Ah i see what you are getting at, so all that throttling stuf you did is a load of rubbish. I have no idea which diode my motherboard reads as the CPU diode, but it indeed is kept pretty much 10 degrees lover than the coretemp temps. Maybe these M0 chips do run hotter than the L2. Aslong as they can take it, thats fine with me. 3300mhz 1.39v, and 72 degrees under orthos, does that temp sound OKish to you? Im still thinking it looks high from my years of keeping cpu temp under 60 on the old amd systems i had.

[GigaByte]

Ah i see what you are getting at, so all that throttling stuf you did is a load of rubbish. I have no idea which diode my motherboard reads as the CPU diode, but it indeed is kept pretty much 10 degrees lover than the coretemp temps. Maybe these M0 chips do run hotter than the L2. Aslong as they can take it, thats fine with me. 3300mhz 1.39v, and 72 degrees under orthos, does that temp sound OKish to you? Im still thinking it looks high from my years of keeping cpu temp under 60 on the old amd systems i had.

Is that 72c on the cores or Tcase (CPU reading)? If its on cores that plenty cool as Tcase is about 62c, if thats on Tcase you are 1c under thermal spec which is still ok.