Realtemp 2.6 can now set manual TJMax! E8XXX 72.4C!

coolchu001

Limp Gawd
Joined
May 27, 2007
Messages
506
I just downloaded the new 2.6 version from TPU and there is now a new setting to set manual TJMax...
Upon further googling on the 45nm TJMax I was surprise to see that Intel actually published the TJMax for the E8400 at only 72.4C!!! THATS MORE THAN 22C lower than the 95C used by realtemp!!

RealTemp 2.6 DL
Intel's Core2Duo specs
Can also be found on page 77 in this E8400 techspecs sheet
More tech specs if interested...

For the Q9xxx....
TJMax for Q9xxx only 71.4C on pg 79

Hope this helps clear some confusion about the E8xxx & Q9xxx series TJMax value...

With a TJMax value of 72C for my Q9450, I am idling at 29C-31C on the cores with a room temperature of 24C (~76F) (This is at stock 2.66GHz with 1.1v (undervoltaged)) load @ 43C-39C on the cores
 
Nice find! Now when I get my E8400 in the next few weeks, I can be sure I have the right temps. Those problems everyone was having was worrying me. Thanks!
 
This is specifying what Tcmax (case max, measured at the center of the IHS) is, NOT Tjmax (junction temperature, measured at the center of the CPU core). Do not set Realtemp's Tjmax to 72.4 and think your CPU is running super cool. Tcmax != Tjmax.
 
Wouldn't the center of the IHS be pretty close to what the temperature is at the center of the core? They're so close, there can't be that much of a difference in temperature.
 
I don't know what the difference is since it probably varies from chip-to-chip, but there is most definitely a difference.
 
So what is the right temperature of a Q9450? Realtemp is showing 58C under load, coretemp is showing 68C. I even reseated my heatsink with no differences in temps. Realtemp with this TJMax of 72 brings the temps down to 26C the room temp was 22C. Does that sound right?
 
scrawnypaleguy, read this wikipedia entry on junction temperature. Junction temps are on the actual core, case temps are on the integrated heatspreader, which is soldered to the core, so there's plenty of layers of materials the heat has to pass through before it gets to the IHS. Here's what it says in the article:

wikipedia said:
In operation it is usually higher than case temperature, the temperature of the part's exterior. The difference is equal to the power dissipated inside the device times the junction-to-case thermal resistance.
So Tcmax is lower than Tjmax by some unspecified amount. Regardless, it's better to play it safe and have Tjmax set at 95C, which seems to be the consensus despite the fact that Intel's never released the actual Tjmax. Besides, setting Tjmax to 72.4C would indicate that my E3110 is IDLING at 10.6C. Sub-ambient temps on air? I don't think so.

NukeULater, there is no way that your Q9450 is loading at 26C. No possible way. The Tjmax is not the same as Tcmax, and from everything I've read Tjmax is likely 95C, which is what Realtemp is reporting. Your chip MIGHT be idling a few degrees above ambient, but I find that unlikely, my E3110 idles at around 30-32C stock and the room temperature is 72F, so about 22 or 23C. This is with a lapped TRUE120 and 2 push/pull fans, so there's no way yours is idling lower than mine when you've got a quad core.
 
Some where I read that you have to add 15c to the TJ temp to get the TJmax making my E8400 a 87.4
 
The TJMax of Q9450 is not 71.4c that is a max core temp the processor can go up too just like 60c limit was set on 65nm cpus. I say the TJmax is 85c or 95c. The 85c seems more accurate to me. But i guess i will leave at 95c. Even with a 95c the temps are not bad for me at 32c idle and 46c load. Bye
 
That's probably a stab-in-the-dark guess.

Oh I am sure someone put like 5 or 10 minutes of thought into that! Most of this is just guesswork and differing theories. I especially like when people deem the TJMax in any given app as wrong because "Now way is my CPU running that hot!" Its too bad people will not standardize on reporting distance to TJMax as their temp :p

Regards,
25C to TJMax Q6600 with 28C ambient, small FFTs
 
This is specifying what Tcmax (case max, measured at the center of the IHS) is, NOT Tjmax (junction temperature, measured at the center of the CPU core). Do not set Realtemp's Tjmax to 72.4 and think your CPU is running super cool. Tcmax != Tjmax.

Why is it that the Tcmax is listed under the Tjmax on the core 2 duo spec sheet? And it appeared in the E8400 tech specs as well. I also went to look for the T7xxx and saw that it says a tjmax of 100C on the tech specs, same as the one on the core 2 duo specs... How can this be explained? The 2 tjmax values (for E8xxx, T7xxx) were both found in the tech spec to have the same value as that listed on the core 2 duo spec page... (72.4C/100C respectively)
 
Why is it that the Tcmax is listed under the Tjmax on the core 2 duo spec sheet? And it appeared in the E8400 tech specs as well. I also went to look for the T7xxx and saw that it says a tjmax of 100C on the tech specs, same as the one on the core 2 duo specs... How can this be explained? The 2 tjmax values (for E8xxx, T7xxx) were both found in the tech spec to have the same value as that listed on the core 2 duo spec page... (72.4C/100C respectively)

Or it could be that those are what is considered the TJMax values when used for an embedded deviced. The Core 2 Duo spec page is for embedded chips information.
 
Why is it that the Tcmax is listed under the Tjmax on the core 2 duo spec sheet? And it appeared in the E8400 tech specs as well. I also went to look for the T7xxx and saw that it says a tjmax of 100C on the tech specs, same as the one on the core 2 duo specs... How can this be explained? The 2 tjmax values (for E8xxx, T7xxx) were both found in the tech spec to have the same value as that listed on the core 2 duo spec page... (72.4C/100C respectively)

It's different because it's labeled as Tcmax, not Tjmax. If it was labeled Tjmax, then that's what the Tjmax would be, but since it's labeled as Tcmax, that's what the Tcmax is, and they are not the same.
 
NukeULater, there is no way that your Q9450 is loading at 26C. No possible way. The Tjmax is not the same as Tcmax, and from everything I've read Tjmax is likely 95C, which is what Realtemp is reporting. Your chip MIGHT be idling a few degrees above ambient, but I find that unlikely, my E3110 idles at around 30-32C stock and the room temperature is 72F, so about 22 or 23C. This is with a lapped TRUE120 and 2 push/pull fans, so there's no way yours is idling lower than mine when you've got a quad core.
Okay room temp was 25C at the time of this screenshot. The CPU is at stock speeds, and Vcore. CPU is running Prime 95, small TTF's.

72 Tcmax
Temps_1.jpg


95 Tcmax
Temps_2.jpg


My previous 3.5GHz Q6700 never even ran that hot, I think something is a little out of whack. I think 72 Tcmax is a bit more realistic temperature wise. Although I might be a little wrong.
 
I don't think Intel has ever published the Tjmax of the E8xxx chips, so any ideas on it are pure speculation. I can guarantee you the Tjmax is NOT 72C. If it is, then my E3110 at 4.05GHz hits 30C at max load at idles at 10C. There's no way that's even remotely correct. My real idle and load temps are 32 and 53, respectively.

NukeULater, what are you using to cool your Q9450? Keep in mind that these things are 45nm and have much higher transistor density than the 65nm because of that.
 
NukeULater, what are you using to cool your Q9450? Keep in mind that these things are 45nm and have much higher transistor density than the 65nm because of that.
I'm running a lapped thermalright ultra 120 extreme, AS5 thermal paste and a 120x38mm panaflo h1 (110CFM at full tilt).
 
That may be causing issues, since the surface on most Intel integrated heatspreaders I've seen has a noticeable valley in the center so now that your TRUE is lapped, it's possibly only making good contact around the outside of the IHS, and not the center, where the cores are.
 
It's different because it's labeled as Tcmax, not Tjmax. If it was labeled Tjmax, then that's what the Tjmax would be, but since it's labeled as Tcmax, that's what the Tcmax is, and they are not the same.

Thank God someone is paying attention.

Dealmaster is exactly correct and the OP is absolutely wrong. Only 1000 posts about this (397 mine) . The OP is to be commended for at least finding and linking the Intel spec. and giving it a good try, it is confusing.

For Tc see the Thermal and Mechanical Guidelines but it is basically the same as what the [H] crew did with cutting a grove and mounting a thermocouple (actually it is exactly the same ). The trouble with Tc is that unless you also cut a grove and mount a thermocouple in the IHS you cannot know what its value is.


/sigh.

Also as mentioned above, delta to Tj already INCLUDES the factory calibration and is all you really need to know. I wish we would start bragging about "Thermal Margin" at idle and load instead of core temps. At least then we would be saying something that means something. Bah I will start by checking now and changing my sig.
 
Thank God someone is paying attention.

Dealmaster is exactly correct and the OP is absolutely wrong. Only 1000 posts about this (397 mine) . The OP is to be commended for at least finding and linking the Intel spec. and giving it a good try, it is confusing.

For Tc see the Thermal and Mechanical Guidelines but it is basically the same as what the [H] crew did with cutting a grove and mounting a thermocouple (actually it is exactly the same ). The trouble with Tc is that unless you also cut a grove and mount a thermocouple in the IHS you cannot know what its value is.


/sigh.

Also as mentioned above, delta to Tj already INCLUDES the factory calibration and is all you really need to know. I wish we would start bragging about "Thermal Margin" at idle and load instead of core temps. At least then we would be saying something that means something. Bah I will start by checking now and changing my sig.

Actually here is a quote from one of the pages he linked:

http://www.intel.com/design/intarch/core2duo/spec.htm

Thats right, it says Tj Max right in the Intel doc and gives a value for it. Now, that may not be the right Tj max, but its definitely saying Tj Max. Guess more than one person is not paying attention.
 
Actually here is a quote from one of the pages he linked:


http://www.intel.com/design/intarch/core2duo/spec.htm

Thats right, it says Tj Max right in the Intel doc and gives a value for it. Now, that may not be the right Tj max, but its definitely saying Tj Max. Guess more than one person is not paying attention.

Right where that value is listed, it specifically states that value is Tcmax, not Tjmax. Look at all the other entries that simply have a temperature. THOSE are Tjmax temps. The ones where it specifies the temperature as Tcmax is the Tcmax temperature, obviously. It's not too hard to understand this.
 
ok I understand that there are differences between tc & tj but why does intel list the tc as its tj temp?? Or simply why did they even not bother to find the real tjmax temps??
 
Actually here is a quote from one of the pages he linked:


http://www.intel.com/design/intarch/core2duo/spec.htm

Thats right, it says Tj Max right in the Intel doc and gives a value for it. Now, that may not be the right Tj max, but its definitely saying Tj Max. Guess more than one person is not paying attention.

And in the box where it lists the value it says "Tc Max = 5° C to 72.4° C" - how hard is it to believe that that means Tc Max? Especially whan all the others just list a temperature value.
 
Some where I read that you have to add 15c to the TJ temp to get the TJmax making my E8400 a 87.4
It all depends on the IC, on the materials, how the IC is made.
I've seen specs anywhere from 6-12C for various ICs.
 
Just a quick note on the e8400 vs 3110 discrepancy that appears to be showing here:
So far it seems the people using the lower TJ have "closer" to real temps than the guy with the 3110 where his chip is obviously not running sub-ambient....it's entirely possible the two chips have different TJ max numbers so you both could be right.
 
ok I understand that there are differences between tc & tj but why does intel list the tc as its tj temp?? Or simply why did they even not bother to find the real tjmax temps??

You'll have to take that up with Intel, I have no idea why they don't publish the Tjmax values. I'm guessing they put the Tcmax values in that same column because there was no Tcmax column and to make the data as concise as possible, they simply put the Tcmax value for the E8400 and the other CPUs in the Tjmax column. Trust me, the Tjmax of these chips is certainly not 72.4C. There is no way.

Just a quick note on the e8400 vs 3110 discrepancy that appears to be showing here:
So far it seems the people using the lower TJ have "closer" to real temps than the guy with the 3110 where his chip is obviously not running sub-ambient....it's entirely possible the two chips have different TJ max numbers so you both could be right.

They don't have different Tjmax values as far as I know. They are functionally and on a circuit level identical. Xeons have some networking capabilities that the regular Core2Duos don't and they tend to be binned a little higher (at least that's popular belief) since their Vid values are lower. They are typically chips that are manufactured slightly better and can run at lower voltages, and hence temperatures, since they are meant to be running 24/7 in a server environment, possibly with passive cooling, so the lowest voltage possible is important.

Trust me, these chips' Tjmax value is not 72.4C, it's likely in the 95-100C range. You can believe otherwise, but there is no evidence to suggest it, nor is there any evidence to suggest that the Tjmax is 72.4C.
 
Another update on the cpu temp readings....currently my room temperature is 17C (around 63F) (AC is on :-D) and the cpu at 2.66GHz at 1.15V idles at 22-26C on each core. Which on avg is 7-8C above the ambient temperature which seems VERY plausible since I am running my Q9450 on a Xigmatek S1823 with a bolt kit and AC MX-2...

These reading are with a tjmax of 73C...
 
Another update on the cpu temp readings....currently my room temperature is 17C (around 63F) (AC is on :-D) and the cpu at 2.66GHz at 1.15V idles at 22-26C on each core. Which on avg is 7-8C above the ambient temperature which seems VERY plausible since I am running my Q9450 on a Xigmatek S1823 with a bolt kit and AC MX-2...

These reading are with a tjmax of 73C...

Dude, are you not listening? The TJmax of those chips is not, not, not 73C.
 
Another update on the cpu temp readings....currently my room temperature is 17C (around 63F) (AC is on :-D) and the cpu at 2.66GHz at 1.15V idles at 22-26C on each core. Which on avg is 7-8C above the ambient temperature which seems VERY plausible since I am running my Q9450 on a Xigmatek S1823 with a bolt kit and AC MX-2...

These reading are with a tjmax of 73C...

I'm convinced you're just being willfully ignorant. The idle temperatures on these chips are known to be quite inaccurate, PLUS Tjmax is NOT 73C. For God's sake, read the spec sheet your are linking to. Look under the Tjmax column with me OK. Following so far?

You'll see that for the T7500, L7500, U7500, T7400, and L7400, under the Tjmax column, there are simply numbers right? That's because those are their Tjmax values. Now, you're going to have to look up again. Look at the E8400, E6400, and E4300. Under their entries in the Tjmax column, what do you see? Is it just numbers?

NO. IT ISN'T. That's because Intel hasn't published the Tjmax values for those CPUs and they might not ever publish them. BUT THEY HAVE PUBLISHED TCMAX. As obviously listed in those entries, it says Tcmax = <some value> for each CPU. Now, if you'd read the wikipedia article I linked earlier, you might have some knowledge on this, but since you're still being ignorant, I'll quote it so you don't have to click on anything:

http://en.wikipedia.org/wiki/Junction_temperature said:
In operation it is usually higher than case temperature, the temperature of the part's exterior. The difference is equal to the power dissipated inside the device times the junction-to-case thermal resistance.

Now, in order for the Tjmax to equal the Tcmax, as you ignorantly think it does, the junction-to-case thermal resistance would have to be ZERO. As soon as you invent a substance that has zero thermal resistance, you let me know, considering it doesn't exist. You are WRONG, WRONG, WRONG and face it. Quit trying to tell us that Tcmax = Tjmax because it doesn't. Those numbers in the Tjmax column are NOT Tjmax values for the chips that have Tcmax listed instead. This really isn't hard to understand.
 
Actually here is a quote from one of the pages he linked:


http://www.intel.com/design/intarch/core2duo/spec.htm

Thats right, it says Tj Max right in the Intel doc and gives a value for it. Now, that may not be the right Tj max, but its definitely saying Tj Max. Guess more than one person is not paying attention.

That is an error in the heading by Intel, they have had that document wrong for 2 years now. You will see in every instance the values listed under the Tjmax in the heading is actually the Tcase when referenced underneath. The fucking heading is wrong and misleading and the document referenced is NOT a technical document btw.

The premise of this entire post is incorrect.


note to Dealmaster: You are more correct than these other noobs but note that even for the mobile processors the Tjmax is given as a range 0-100C. A little thought would indicate that could not be right. Tjmax is a single value where PROCHOT is tripped. It cant be a range. Those are the recommended working temps or some such crap similar to Tcase but I have forgotten where they came from. The mobiles actually had the Tjmax in a register or there was an MSR register you could get a bit value out of to tell you if the Tjmax was 105C or some other value if I recall correctly, it was over 2 years ago so I dont remember exactly. Other than it caused havoc when people tried using the same code routine to "read" the Tjmax of the very first C2Ds and it did not work because Intel clearly states:

The Tjmax is set at the factory for each cpu and it varies.

This alone is sufficient reason to realize why Intel cannot publish a "hard" number for a series or family of cpus. It does not exist.

Here is the classic by Uncleweb
http://hardforum.com/showpost.php?p=1031827510&postcount=31

Here is one where the same mistake as was made by this posts OP by attempting to read marketing/sales info as technical data.
http://hardforum.com/showthread.php?p=1031827510#post1031827510


I am done, its not worth my time. Its a free country think what you want.
 
The TJmax is 95c on Q9450 the 72.4c is the max core temperature it can go up too its Not a tjmax. Now someone might say the TJMax is 60c on 65nm cpus too. When in reality its the core temperature and on 65nm the TJmax was 85c for Dual core.
 
Back
Top