C2D temps revisited....

And just for grins, here are my LOADED temps at 1600mhz and 1.2v (set in BIOS, vdroop not considered). For those not aware, coretemp and TAT do not read cpu speeds correctly when you change the multiplier to a different setting than default. The actual speed IS indeed 266x6 for 1600mhz. cpu-z gets this right.

loaded_wtf_temp.jpg
 
Thanks TheRapture for taking the time to prove this as best we can.

By reducing our cpus down to 1600 MHz and approximately 1.17 volts we both ended up at about 2C or 3C above ambient. This would mean that the Tjunction temp for all desktop processors is 85C. The only exception so far is the E4300 which during similar testing proved that it couldn't be 85C so was assumed to be 100C.

If you have an E4300 use CoreTemp 0.95. If you have a revision L2 E6300/E6400 as reported by CPUz then use CoreTemp 0.94 which will report it with the correct 85C Tjunction.

I really hope as many people as possible can repeat this test. It's simple to do and the more data we have the better.

Edit: Here's Orthos at 1600 MHz and low volts. 34C isn't too shabby with the OEM heatsink and fan! :D
 
dook43 said:
I'd say that's conclusive evidence in our favor. I'd say a 2.8Ghz overclock on stock cooling at 56/57 load ain't bad. Think I can go for broke with 3.2? :D :D :D
Click to expand...


If you have good cooling and keep the vcore under ~1.45v, shit, 3.2ghz should be a good target. I am back at 3200 right now, load testing with vcore set in bios to 1.45 which gets me a real reading of 1.42v idle and 1.39v under load. Vdroop on the DS3 is always about .03 LESS than what you set in bios.
 
Here is a pic of the current setup. Notice that speedfan 4.32 is at 65c load, and coretemp .95 is 80c. Exactly 15c difference. Don't know about TAT since it is not designed for desktop cpus I think, it is splitting the difference at only 10c or so higher than speedfan :D

3200_testing.jpg
 
TheRapture said:
There has been much lengthy discussion on the C2D temps using TAT, speedfan 4.32, and coretemp .95....

So, which one is correct? I have the E6400 (Allendale) with a big tyhpoon, at 3.2ghz and 1.36vcore under load, which gets me either 60c load temps (speedfan) under ORTHOS or 75c load temps (TAT and Coretemp agree)?????

One of them is wrong....while I would like to believe the lower range I want to be sure, since TAT loading can push my temps up to 80c or so....
Click to expand...

TheRapture--

See my post from another thread in the Intel MOBO forum--MSI P6N SLI thread--..... There are a couple of links in it that got me started in believing that both TAT and Coretemp 0.95 are NOT RIGHT! They are indeed 15degC too high and this is not because of a the actual temperature measurement but an error in identifying the newer stepping on the 63/6400 Conroes (you have L2 like mine). The heuristic (it's a calculation not a direct measurement) then that the program uses is then thrown off and you can see the difference when you load/run Coretemp 0.94-- which corresponds to Speedfan.

3degree said:
I'm getting these high temps too....

OK, I am pulling together a bunch of pieces of posts trying to make sense of what I've got going on and what may be happening. AND looking for help :(

So here's the story and where this is heading.....

To start it off, I got in a 6400 and an Asus P5B Dlx and put a Zalman 9700 on it. At stock, My idle temps were in the high 30's and at load into the high 50's. With an OC to 2.8-3.2GHZ, my temps went up over 65DegC from orthos and I cut off the test before 70degC. I thought that the Zalman was not right (mounting, Zalman thermal paste, something.) I reinstalled and with Artic Silver and saw little to no difference. These temps were nowhere in line with what I see out on the Asus P5B threads from other Zalman users. I tried to put on the stock HSF to see if it made a difference. Somewhere in the reinstall, the MOBO/CPU borked and the board failed to even post or give error beeps, etc. no matter what I tried. (and this from a simple HSF install???) I returned the board and switched to the P6N platinum (as there were other problems with that board that were driving me nuts) and RMA'ed the CPU to get another 6400, and sent back the Zalman for a Tuniq.

I did my OC this weekend. At stock with a Tuniq Tower 120, I'm getting idle temps at ~ 38-39 degC and loads in the high 50's!!! I went through from 1333, 1400, and 1500 settings with some bumps in both Vcore and NB voltage. I got it 4hr+ stable on orthos-small tests and posted booted into Win fine, running 3Dmark 05 and 06 and getting numbers at each level. So I can run 3.0Ghz w/o a problem- no errors, stable, etc. [It would post at 1600 but die in WINXP. And at 1640, it failed to post (and I had the Vcore at 1.48, NB 1.4, FTT 4%). But I was worried I should stop there anyway.......]

The problem is my temps. :eek: At idle, they were still/stayed at ~39-40degC no matter what I threw at it. This seems pretty high for the Tuniq from what I've read, and from some of what I've seen here. I'm taking the temps from Coretemp 0.95 and TAT. But have also checked against speedfan and Everest too. (big differences).

At load, I was getting up to 65degC, at the last with running all day. This is still really high for the Tuniq. (almost unbelievably so). My case is a P180B and I've got decent airflow; my case temps are good via speedfan or the bios monitoring. [It was cool in the room yesterday ~65F with the windows open.] So I opened up the case and when I touch the HSF on the base, it is only very slightly warm when it's SUPPOSED to be ~65C according to Coretemp .95 and TAT. :confused: The heatpipes are cool, and the NB is cool (got the optional fan on it spinning away at 8000 rpm).

So, on a search I come across these posts:

http://forums.extremeoverclocking.com/t254892.html

http://www.alcpu.com/forums/viewtopic.php?p=980&sid=f54fa3e3da9ebcede609f279ec1f13eb

http://forumz.tomshardware.com/hardware/C2D-E4300-GA965P-DS3-v3-F10-temps-ftopict231432.html (I know it's for a diff MOB, but the temp programs are the same).

Don't know if you guys have seen these..... I can't seem to verify the idea, but it seems to make sense and fit what I've seen in that I've had 2 different 6400's with very different (but still top notch) HSF's on them and am running hot no matter what- at least according to Coretemp and TAT. The posters from the other forums are making the case that the calculations done by these programs may be wrong--at least for a misidentified stepping on the newer Core2duo 63/400's

I'm going to find a download of 0.94 of Coretemp and see if that helps as proposed, and will have to look at the Tjunction and Tmax indicators they mention when I get home tonight from work.

Can you guys (who are running warmer than expected) check and list your steppings- to see if these are indeed L2? and see if yours are running at 85 or 100?
Click to expand...


I also ran across a thread at extreme overclockers forums where these guys were having a similar issue with unusually high heats on these software tools--- and it was even more unbelievable as they were on watercooling!

Gotta go with what's logical and make sense in the real world. My HSF should be warm to the touch if I'm getting those kind of temps and it's not!

I've been told that --well, those temps have to be true because they are not out of line with what other people get-- and yeah it looks to be verified in the MSI P6N thread here that all of us with high heats (unusually so) all have the L2 stepping Conroes and are about 15deg too high on both TAT and Coretemp 0.95. 15degC lower (and more in line with what other people (with B2 stepping Conroes) get with our coolers and cases). Go figure.
 
msg1285 said:
I have the same almost the same setup as you and I'm getting the same results TAT is lower than Speedfan/Coretemp.

I think that until we can find out for sure which chipsets TAT supports this is kind of pointless. It seems to me that if all three programs are supposed to be reading the same sensor then something has to be throwing them off.[/QUOTE]

You're very right! It is. Look at the Tjunction difference in 0.94 vs 0.95.
Click to expand...
 
I see the exact same thing on my e6400 L2 revision. I get temps that are 15 degrees higher in coretemp 95 than I get in speedfan or coretemp 94.

I do not have the best cooler for my C2D but it is not the worst either, the Zalman 7000 should be a lot better than the stock cooler. If coretemp 95 is right, than my Zalman is doing a terrible job cooling the e6400.

Right now it seems that speedfan or coretemp 94 are both much more believable on my setup.
 
Well I'd have to agree.

I got my overclocked rig (see Sig) running under load at around 68c or so. I'd been playing wow like 6 or 7 hours straight one weekend (I don't do that a lot at all) and no problems. So I put Core Temp back on my computer (had to remove it for some reason) and it crashes XP every time I open it. So I loaded it in my Vista partition and with Orthos running it hovers in the 85c range!!!!!!! :eek: Orthos used to run a max load temp of around 68c before....

When I move it back to stock it is running like 71c under load (with stock speed and HSF) I've got it on right.. but it seems this new version of Core Temp has the same reading as TAT and TAT was way too high for me earlier. Yeah I think there is something fishy going on. Is there a link to the old version anywhere. I'm at work and can't search for it.

If not no biggie.
 
The problem is that Tjunction or TjMax is not properly documented by Intel so individual software developers can not write temperature monitoring programs that are 100% accurate across their entire line of processors. It's a guessing game and the only company to blame is Intel.

The calculation is simple:

core temperature = Tjunction - DTS

Intel documents how to read the digital thermal sensor ( DTS ) value but does not document how to read the Tjunction value which is needed. Without knowing the correct Tjunction it is impossible to calculate an absolute core temperature value. One person from Intel implied that there is no way to read Tjunction from a processor which may or may not be true.

An assumed Tjunction of 85C for the E4300 results in idle temps below ambient which is impossible so it is assumed to be 100C.

Unfortunately the developer of CoreTemp also assumed that the new E6300 / E6400 cpus built on the same Allendale core must also have a Tjunction of 100C but TheRapture proved that to be wrong in his above post.

Here's what seems to be true:

Tjunction:
E4300 = 100C

All other presently available Core 2 desktop processors have a Tjunction of 85C. This includes the original revision B2 Conroe based processors as well as the E6300 / E6400 revision L2 processors.

Laptop processors have a Tjunction of 100C. This includes all Core Solo, Cord Duo and Core 2 Duo mobile processors:
T1300, T1400, T2300 to T2700, T5500, T5600, T7200, T7400, T7600

CoreTemp 0.95 works properly on all processors except the revision L2 E6300 / E6400.

If you have a revision L2 E6300 / E6400 then use CoreTemp 0.94.
 
unclewebb said:
CoreTemp 0.95[/URL] works properly on all processors except the revision L2 E6300 / E6400.

If you have a revision L2 E6300 / E6400 then use CoreTemp 0.94.
Click to expand...

Man you guys gave me a huge relief. Thanks for all your work on this. I have one of the offending L2 E6300s with the 15C variation between CT 0.94 and CT 0.95. Everest, SpeedFan, and CT 0.94 generally agree with each other. My machine is in a very cool basement. I have a ninja plus cooler and lots of fans. It seemed crazy to me that my temps were reaching 60C+ in Orthos with my setup. Even subjectively it was weird, cause I'd feel the case fan output by my CPU and it was blowing cool air.

Looks like temps were more like mid-40s. I have only slightly bumped CPU, NB, and RAM voltages (all +0.05).

Thanks again - I'm sticking with CT 0.94 til I hear otherwise. This is a great forum.
 
I was trying to hold my voltages down a bit, thinking my temps were borderline, but now that I am sure I was way in the green, I went ahead and bumped it up to finally stabilize my overclock, and man, it has been rock solid. In the BIOS of my DS3 rev. 3.3. vcore is set to 1.45 (yes my Allendale needed it), up from my previous 1.425, and this thing has done 9+ hours of ORTHOS, loops of 3DMark06, and two instances of Folding@Home going....stable enough for me. At the previous 1.425 vcore setting in bios, if I left ORTHOS on long enough (4+ hours) it would crash, and even running DreamScene overnight would eventually crash the rig as well. Now it has not even burped since raising the vcore to 1.45.


Even in my somewhat warm room, typically 27c-28c ambient, running ORTHOS and dual Folding@Home cores, my cpu under ORTHOS is not breaking 67c now....a bit warm but hey, it's at 3200mhz with a real vcore under load of 1.39v. Still well under the safety limits and I won't even concern myself with temps now unless I see something approach 70c, which it never will with current gaming loads.

finaloc3200.jpg
 
TheRapture: I've found that instability can sometimes be caused by a lack of memory voltage.

Everyone keeps telling me the Allendale core cpus need more voltage but I'm not an easy guy to convince. :D

If you're ever curious try setting your cpu core back to 1.425 volts and bump up the memory voltage to 2.1 - 2.2 volts. If you can get your system as stable as it is now while using less cpu volts you might have room to overclock some more.

The first E6300 I built with a Conroe core was good for 3300 MHz ( 472 x 7 ) and I think I only had it set to 1.35 volts in the bios. The DS3 was an incredibly stable motherboard at any speed.
 
My ram is rated (taken with a grain of salt here) at 1.8v, since I am not overclocking it any (400x8) I set it in the bios to +.2v which gets me 1.98v as per speedfan. Do you think bumping it up voltage wise would do any good considering it is running at stock rated speeds?
 
I go by the theory that it never hurts to try.

I'm presently just using some cheap Crucial ram that was factory rated at 1.8 volts but it seems to like 2.1 - 2.2 volts. I haven't pushed it too far. The Micron D9 based memory modules like that much voltage and more as shown in this review:
http://eclipseoc.com/index.php?id=1,72,0,0,1,0

They really scale nicely.

If it allows you to reduce your cpu voltage and overclock higher it might be worth it but there's nothing wrong with 400 x 8. You lose a lot of memory bandwidth the moment you go 1 MHz higher than 400 so it might not be worth it for you to go only slightly higher than 3200 MHz.
 
Rapture: and what does TAT say running both cores @ 100% load? Your temps are pretty insane. With a 3GHz allendale core cpu and coretemp 0.94 I get 53c max on both cores running orthos small fft. And besides those 0.94 temps are pretty bogus. Use 0.95 to get a better picture on actual temps. :)
 
SuperKeijo said:
And besides those 0.94 temps are pretty bogus. Use 0.95 to get a better picture on actual temps. :)
Click to expand...

For the L2 revision Allendale E6400 and 6300, Coretemp .95 is WRONG. TAT shows me at 74c or so under load, that is wrong as well.
Speedfan 4.32 and Coretemp .94 are reporting correctly for these cpu's.

For the E4300, .95 is CORRECT while .94 reports WAY too low. Read up this page and onto the previous page 2. We have proven which version works correctly with which cpu. My temps are fine and well within safe limits.

TAT is wrong on many cpu's as it was designed for mobile cpu's. To sum it up as unclewebb stated above:


All other presently available Core 2 desktop processors have a Tjunction of 85C. This includes the original revision B2 Conroe based processors as well as the E6300 / E6400 revision L2 processors.

Laptop processors have a Tjunction of 100C. This includes all Core Solo, Cord Duo and Core 2 Duo mobile processors:
T1300, T1400, T2300 to T2700, T5500, T5600, T7200, T7400, T7600

CoreTemp 0.95 works properly on all processors except the revision L2 E6300 / E6400.

If you have a revision L2 E6300 / E6400 then use CoreTemp 0.94.
 
TheRapture said:
For the L2 revision Allendale E6400 and 6300, Coretemp .95 is WRONG. TAT shows me at 74c or so under load, that is wrong as well.
Speedfan 4.32 and Coretemp .94 are reporting correctly for these cpu's.

For the E4300, .95 is CORRECT while .94 reports WAY too low. Read up this page and onto the previous page 2. We have proven which version works correctly with which cpu. My temps are fine and well within safe limits.

TAT is wrong on many cpu's as it was designed for mobile cpu's. To sum it up as unclewebb stated above:


All other presently available Core 2 desktop processors have a Tjunction of 85C. This includes the original revision B2 Conroe based processors as well as the E6300 / E6400 revision L2 processors.

Laptop processors have a Tjunction of 100C. This includes all Core Solo, Cord Duo and Core 2 Duo mobile processors:
T1300, T1400, T2300 to T2700, T5500, T5600, T7200, T7400, T7600

CoreTemp 0.95 works properly on all processors except the revision L2 E6300 / E6400.

If you have a revision L2 E6300 / E6400 then use CoreTemp 0.94.
Click to expand...


And this is based on what? What makes you say that coretemp 0.95 is right for e4300 but wrong for L2 6300/6400? From my understanding an allendale cpu is an allendale cpu, why would e4300 have a tjunction of 100 degrees while the others 85? That just doesn't add up. I believe all allendales have the same tjunction, wether it's 100 or 85, I'm not sure about. Because what you're saying would meen that e6300:s run 15 degrees cooler than the e4300, which isn't true.

I want to see some facts that e4300 and L2 6300s have different tjunction, please.

EDIT. just to make things clear, I have an e4300 cpu. I've done a little testing with it and coretemp 0.95 and 0.94 on idle temps. Running it at 3,2GHz and 1.45v IDLE temps are ~46 on 0.95 and ~30 on 0.94. With 3GHz and 1.35v temps are 42-44 and ~28. And running the cpu stock 1.8GHz and 0.9v temps are 38-40 on 0.95 and ~25-26 on 0.94. At 1.2GHz and 0.7v temps are still the same. Room temp is ~24c. So you're telling me that an e4300 at 1.2GHz and 0.7v running idle on desktop doing nothing at 40 degrees celsius is the correct reading? That is with a Ninja heatsink and a 12cm fan running at 12v (1200rpm). The heatsink is just as cold as any other metallic item in the room. And the ninja is properly contacted to the cpu. Gets a very good contact and has a nice thin layer of thermal paste between the two. The heatsink was lapped (I bought it second hand) but the cpu isn't. I can run 3dmark tests at 3,42GHz and it doesn't crash (needs a shitload of voltage though), so the contact isn't a problem. So me running a e4300 at 1.2GHz and 0.7v idle temp of 40 is the right reading, and you running a 3ghz+ at under 30 is correct?

EDIT 2. I once even tried running TAT @ 100% with the case fan and cpu fan off. At 3.2GHz and 1.45v. I let the temps go up to 96-98 (acording to TAT) and it didn't crash on me. Then I put the fan back on and all was well once again..
 
Which REVISION of the E4300 do you have? That makes the difference. What I AM saying is that my L2 revision, Allendale, E6400 cpu, that Coretemp .95 and TAT are WRONG. I did not say that you have the same problem. Did you even read the tests we did? If your cpu IDLE temp is LOWER than your AMBIENT temp, then that reporting software is WRONG, since with aircooling you cannot have that happen.

I do not know your revision....it makes all the difference. Like I said, did you read the entire thread.


EDIT: it appears that with your cpu, Coretemp .94 is indeed reporting correctly IF your ambient temp is ~3c lower than your idle temp. Same as my setup. I get idle of 27c to 30c depending on my ambient temp temp with the cpu at 1.45v and 3200mhz. Spot on. Coretemp .95 shows me at 44c-47c under the same conditions, which is flat out wrong.
 
There seem to be some changes in the current crop of Allendales and E4300's, the main idea is, one or the other is correct (coretemp .94 or .95).

Basically, idle temps should be between 3c and 5c higher than the ambient temp on a well cooled cpu. If the particular program reads LOWER than ambient, that program is wrong and the other should be used. In my case, Coretemp .94 and Speedfan 4.32 match perfectly while TAT reads about 10c higher, and Coretemp .95 is EXACTLY 15c higher. TAT I will not trust at all since it was not designed to read off of these newer desktop cpu's, it was originally for mobile chips which usually have a TJunction of 100c.


Sorry if my other post was rude sounding? I did not mean it to be so....
 
unclewebb said:
I'm certain that the E4300 has a 100C Tjunction temperature and is being reported correctly by CoreTemp 0.95. I don't yet have any evidence concerning the new Allendale E6300/E6400.

I hope someone with an E6300/E6400 Allendale can use my testing method to try and confirm what the Tjunction actually is. CPUz reports these new processors as Revision L2. Ignore other software programs that sometimes get the Allendale / Conroe info wrong.

dook43: My test is to slow the processor down and to drop the voltage so it produces as little heat as possible.

SpeedFan v.4.32 reports the temp of my Revision B2 Conroe core E6400 the same as CoreTemp 0.94 or CoreTemp 0.95 so I used SpeedFan to draw a graph.

I started with 6 X 333.3 = 2000 MHz with the core voltage at 1.328 volts which is as close to the Intel default of 1.325 volts as I could get. Idle temp at this setting with the OEM cooler was between 29C and 30C.


When using SpeedStep I have watched CPUz report a lower voltage on my board but I think it is an error. The only way to be guaranteed is to lock the multiplier and the voltage in the bios.

Next I dropped the bios down to 6 X 266.6 = 1600 MHz. This allowed me to drop the core voltage to 1.200 volts in the Asus bios which gave me 1.176 volts as reported by CPUz. Idle temp dropped down to 24C. I then opened up the case and got the idle temp down to 21C with the occasional drop to 20C. Room temperature was 18C to 19C.

E6300 or E6400 and ends up with idle temps below their ambient temp then we will know that the software they are using is misreporting the core temperature.

This testing method is a lot more friendly to your processor than bringing it up to 85C or 100C.
Click to expand...



This is the method I used to determine that my cpu is correct using Coretemp .94 and Speedfan 4.32. The temp deltas indicate these are correct and jive with other users with the same revisions and some common sense. No way my cpu was idling at 45c in a 25c ambient room (which is what TAT and Coretemp .95 said).
 
This is one big confusing mess that Intel has created by not properly documenting Tjunction and how to read that value from a processor.

My conclusion that the Tjunction is 100C for the E4300 was because of several users reporting that their core temperature as reported by CoreTemp 0.94 was too low. CoreTemp 0.94 used a Tjunction of 85C for these processors but that resulted in temperatures that were less than ambient during idle which is impossible.

SuperKeijo: How about posting a screen shot of SpeedFan or CoreTemp 0.94 or CoreTemp 0.95 with your processor locked at 200 x 6 in the bios and the voltage set to 1.20 volts. Also include a CPUz screen shot since that seems to report voltage accurately at this level.

We'll try not to jump to any more conclusions. For TheRapture it seems that his Tjunction is 85C with his E6400, revision L2.

If I get enough data maybe I'll take a stab at writing my own temperature utility program. Research is still in progress. :rolleyes:
 
unclewebb said:
SuperKeijo: How about posting a screen shot of SpeedFan or CoreTemp 0.94 or CoreTemp 0.95 with your processor locked at 200 x 6 in the bios and the voltage set to 1.20 volts. Also include a CPUz screen shot since that seems to report voltage accurately at this level.
Click to expand...

Sure, I'll do this. But at those frequencies I can run it at 0.7v, wouldn't that be better? to run it at as low voltages as possible?
 
SuperKeijo said:
Sure, I'll do this. But at those frequencies I can run it at 0.7v, wouldn't that be better? to run it at as low voltages as possible?
Click to expand...
I was hoping you could run it at the same voltage as TheRapture and I did for consistency. You should be within a degree or two of absolute minimum that your cpu can run at so it shouldn't make much of a difference as your previous testing showed. Remember to include your room temperature and any other relevant details.
 
unclewebb said:
I was hoping you could run it at the same voltage as TheRapture and I did for consistency. You should be within a degree or two of absolute minimum that your cpu can run at so it shouldn't make much of a difference as your previous testing showed. Remember to include your room temperature and any other relevant details.
Click to expand...

Ok, I'll get to this tomorrow.
 
I guess I got to it now. Exactly 15 degrees difference between 0.94 and 0.95 on both cores. TAT and 0.95 same temps..

Here's my input:

temps.JPG


Ambient temp was 21.3 degrees this time..
 
At 6 x 200 and 1.2 volts your cpu temperature at idle should start to approach your room temperature. Usually you get within 2C to 4C of ambient room temperature with your case open or if you have good case air flow.

SuperKeijo: Your results show that your Tjunction must be 85C which means that CoreTemp 0.94 is the one to believe for your E4300.

I wish this cleared things up but it only leads to more confusion. Some E4300 cpus running this exact same test have proven that their Tjunction must be 100C and yours has proven the opposite.

Is it possible that Intel changed the Tjunction of the E4300 sometime during manufacture? I guess anything is possible.

Do you still have your CPU box which shows the part number and date code? The early E4300 processors had a Tjunction of 100C. The first date code for them is Q640 which corresponds to a manufacture date of the 40th week of 2006.
 
SuperKeijo said:
Yeah I have the box and it says Q644. I guess that's quite an early e4300 then?
Click to expand...
In the [H] OC database the E4300 ranges between Q640, the earliest, to Q644, the latest.

They seem to manufacture a batch of processors for a month and then switch to producing something else.

Q644 seems to be a late version of the first batch of E4300 cpus if that makes any sense.
 
Hmm

I have an e6600, L639A659 from the box

TAT: 41c under full load (@ 3ghz)
CoreTemp .95: 41c under full load
CoreTemp .94: 41c under full load (crashed when I closed it first time?)

CoreTemp/Cpu Z show it as Rev B2.

So looks like each of those programs are identical with that batch of e6600s.
 
jmackay: The problem isn't with the Conroe revision B2 core processors like your E6600. All of them have a Tjunction of 85C so SpeedFan 4.32, CoreTemp 0.94 and 0.95 will all report the correct core temperature. TAT is usually within a couple of degrees of those other programs at full load.

DTS is the digital thermal sensor temperature and Intel has documented how to read that value. After a program reads the DTS from each core of a C2D it uses a simple formula to convert that number to a core temperature:

core temperature = Tjunction - DTS

The problem is with the Allendale core E4300, E6300 and E6400 processors. These are reported as revision L2 by CPUz. There seems to be some E4300 processors with a Tjunction of 85C and some have a Tjunction of 100C. Intel has not documented how to read the Tjunction value from a processor or even if it's possible to read this value. Without knowing Tjunction, you can't properly calculate the core temperature.

It would be great if more people with one of these L2 processors would run the same test as we did above to try and determine what their Tjunction really is. Even people with a B2 core Conroe running the above test will help confirm that this is a valid way to test if your temperature program is working properly.
 
jmackay said:
Hmm

I have an e6600, L639A659 from the box

TAT: 41c under full load (@ 3ghz)
CoreTemp .95: 41c under full load
CoreTemp .94: 41c under full load (crashed when I closed it first time?)

CoreTemp/Cpu Z show it as Rev B2.

So looks like each of those programs are identical with that batch of e6600s.
Click to expand...



Holy crap even more ambiguity. :p

The same base principle applies. Lower your speed to a very low level at a reduced voltage, check idle temps, and monitor temps. Whatever program that reports 2c-5c above your ambient temps, it most likely correct.

EDIT: ahh, I see, the real Conroes don't have this issue...I forgot, it seems to be only the Allendales.
 
unclewebb said:
It would be great if more people with one of these L2 processors would run the same test as we did above to try and determine what their Tjunction really is. Even people with a B2 core Conroe running the above test will help confirm that this is a valid way to test if your temperature program is working properly.
Click to expand...


Yep, come on people, it is an easy test to do. Crank that sucker down to the same speed and voltage as all of us have tried, and let's cement this theory into the ground.
 
unclewebb said:
jmackay: The problem isn't with the Conroe revision B2 core processors like your E6600. All of them have a Tjunction of 85C so SpeedFan 4.32, CoreTemp 0.94 and 0.95 will all report the correct core temperature. TAT is usually within a couple of degrees of those other programs at full load.

DTS is the digital thermal sensor temperature and Intel has documented how to read that value. After a program reads the DTS from each core of a C2D it uses a simple formula to convert that number to a core temperature:

core temperature = Tjunction - DTS

The problem is with the Allendale core E4300, E6300 and E6400 processors. These are reported as revision L2 by CPUz. There seems to be some E4300 processors with a Tjunction of 85C and some have a Tjunction of 100C. Intel has not documented how to read the Tjunction value from a processor or even if it's possible to read this value. Without knowing Tjunction, you can't properly calculate the core temperature.

It would be great if more people with one of these L2 processors would run the same test as we did above to try and determine what their Tjunction really is. Even people with a B2 core Conroe running the above test will help confirm that this is a valid way to test if your temperature program is working properly.
Click to expand...

Oh I know what you guys were doing, just confirming that all programs work correctly for my proc / mobo (Evga 680i A1)
 
TheRapture said:
Yep, come on people, it is an easy test to do. Crank that sucker down to the same speed and voltage as all of us have tried, and let's cement this theory into the ground.
Click to expand...

I would love to do that and help out..... but I cannot seem to change the CPU multi. Like right now I have it set at 333*6 but it is being reported by CPU-Z and everything else as 333*9. I dunno if it is my CPU or my mobo (or BIOS) that is doing this crap to me but I wanna run 375*8 instead of 333*9. There is a BETA BIOS out for my mobo. I am beginning to think that this is a craptacular mobo (Abit IB9) and that Abit isn't going to support it anymore as far as the BIOS goes. I have no other beef's with the motherboard other than the multi thing that I am trying to figure out right now.

FWIW: Coretemp 0.94 22c-23c idle
Coretemp 0.95 37c-38c idle
Everest 4.00.976 22c-23c idle
TAT = NO WORK

333*9 @ 1.350vcore
 
Ok, these are the results of my E6400 revision L2 with the multiplier set to 6X and the voltage dropped in the bios to 1.175v. In speedfan the the voltage shows 1.14v and in cpuz it says 1.136v. Tempatures in the house are around 18c-20c and I have the case open,

 
CpuMan: Is there any way you can get a more accurate room temperature reading, preferably closer to the fan on your cpu? I know in my room, the temperature displayed on the digital thermostat can be a degree or two different from the air entering my computer at the floor level.

The processors with an actual Tjunction of 100C that CoreTemp 0.94 was using a Tjunction of 85C to calculate the core temperature were generally about 10C below room temperature during this test making it very obvious that CoreTemp 0.94 was very wrong.

With your data I tend to believe CoreTemp 0.94 and SpeedFan 4.32 which both use a Tjunction of 85C as the one you should be believing.

By the way, I suggest you close your computer case back up.
If one of your fans that is running at 1,350,000 rpm ever gets loose it's going to take someone's head off! :D
 
cpuman....nice and cold in your room eh? Man if my room was that chilly.....yikes. 25c to 27c here, but then again, it's in south texas, and already the days are getting to the high 80f range....summer here sees 98f-100f alot of the time and keeping a house any cooler than ~75f gets to be expensive and the damn unit runs all the time. :p

Looks like what unclewebb says is right.
 
Well I am goin to do some testin tonight for you guys. I managed to get the 1.3Beta BIOS flashed into my motherboard and now I am able to change the cpu multi and it sticks all the way into windows.
 
You must log in or register to reply here.
Back
Top