• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Tjunction?

spartus4

Limp Gawd
Joined
Apr 6, 2004
Messages
139
What is the Tjunction? When I run CoreTemp it says that it is 100C. Is this normal? The processor that is in the system is a Q6600 G0. I am just wondering if someone could explain what the Tjunction is and how it reflects on performance?

Thanks

spartus4
 
Hiyas ( cheers, first post on [H] ) ..

What you're asking is actually perfectly explained on CoreTemps homepage ( http://www.thecoolest.zerobrains.com/CoreTemp/ ) Its currently down though, i assume from the looks on http://www.thecoolest.zerobrains.com/ that he reinstalled hes web server ;)

Anyways, AFAIK it does not have any effect what so ever on performance. I remember asking myself the exact same question few weeks ago when i had my own q6600, and i read it on that particular homepage, so unless someone pops up with a fulfilling reply, its waiting just around the corner when the CoreTemp web page is back up ;)
 
OK, fasten you seat belt.

Nothing to do with performance.

You cannot damage a C2D with heat, it will just shut off, cool down, and should be good to go. You can damage them by too much voltage.

Based on the Intel data sheets that are public information (see references at bottom) and Intel white papers (discussions of technology by Intel designers) this is what I understand it to be.

Intel processors have a DTS (digital temp sensor) built into the chip. There are actually several sensors built into the CPU transistor circuitry at several different places in the CPU die in an attempt to cover hot spots. So the DTS is a system not just one sensor and the DTS will poll these different sensors and make the raw data corresponding to the highest temp available in a CPU register we can read with software.

Additionally there is a TCC (thermal control circuit) that takes action when the DTS reaches a preset point/temp.

Tjunction is not referenced in any Intel document I have access too. A discussion on Intel's own forums seems to indicate this is an old term. Classically it was used to denote the junction temp of a transistor. I believe this term was rolled over in early thermal monitoring systems to indicate the maximum allowed internal CPU temp. There is support for this theory in the Intel forums where an Intel Engineer mentions it would probably be more accurate to call this value Tjunction-max.

What this is (Tjunction) is a temp that Intel, through experimentation and experience with the process used to make the CPUs has determined as the maximum safe operating temp. There is a second temp that is referred to as the Catastrophic Trip temp at which the CPU will shut down.

The way it works is this:
The DTS monitors internal temps and reports them as a negative value that corresponds to the number of degrees C BELOW Tjunction. Think of it as the "headroom" or a "countdown" until action is taken. If the processor's temp continues to rise this value becomes closer to 0 which corresponds to the Tjunction trip point being reached, in your case 100 C. If the DTS reported temp matches Tjunction the value in the register reads 0 and the CPU will send out a PROCHOT# interrupt signal. PROCHOT# can be thought of as the CPU screaming at the top of its lungs "I am too hot, do something quick or I am outta here!!! " This PROCHOT# can be used by board designers any way they want (alarms etc) Additionally the TCC circuit will kick in to take action to cool the processor down. If EIST is enabled in the bios, fans will spin up to 100% speed, TM2 (thermal monitor 2) if enabled in the bios, will attempt to lower the CPU multiplier to reduce CPU speed and thus heat, CPU core voltage will also be reduced. C1E will "steal" clock pulses in an attempt reduce the workload on the CPU. But lets say all that stuff is disabled in the bios cause we are leet overclockers and also we where too lazy to hook up a buzzer to the motherboards front panel pin connectors and our CPU fan just failed. The CPU will continue to heat up and at a few degrees above Tjunction the Catastrophic trip temp is reached. This temp is not public data and I am not going to speculate. Anyway the CPU hits Catastrophic trip and the CPU shuts down to save itself and your machine crashes because the CPU just turned itself off.

Thats how it works, so what does this mean ?

Tjunction is different for different families of CPUs, quads vs duals vs ? Unless you have an inside developer relationship with Intel you cannot tell what the Tjunction really is. When the C2Ds first came out there was a LOT of experimentation when the coretemp application first came out. At the time it was thought you could find/read the factory programmed Tjunction trip point in another CPU register. This proved to be false. So sometimes the software (coretemp and later others) had Tjunction incorrect, especially when people tried to use it with older processors or processors just released and the developer had not had time to update the program. Long story, lots of misunderstanding, but it was fairly quickly sorted out that a Tjunction of 85C was correct, or close enough for dual cores, 100C for quads and the new just released chips have a Tjunction of 105C. So the version of coretemp you use matters.

So that 100C you see is the best guess at what the maximum recommended internal temp should be allowed to rise to during normal operation. The software just does the math for you because as I mentioned the only thing we can get for damn sure is the "delta" in deg C to the "I am too hot" trip point (Tjunction). So with a quad it is assumed Tjunction is 100C and the program reads the register to determine the delta and does the simple 100C - delta = actual core temp and spits it out as your coretemp.

You can read this value yourself if you like, here is how, its pretty easy:
http://www.hardforum.com/showpost.php?p=1031080147&postcount=5



Now, are you not glad you asked? :D

http://www.intel.com/technology/itj/2006/volume10issue02/vol10_iss02.pdf
http://www.intel.com/design/core2duo/documentation.htm
(The Thermal and Mechanical Guidelines document and the datasheets themselves. )
http://www.xtremesystems.org/forums/showthread.php?t=167446
 
Quite interesting. From what I understand of your explanation, around 100C would be a point where a Q6600 would go into crazy shut me down mode? I would have thought it would have died way way before 100C.
 
It is pretty amazing. yep for a quad in theory you could run it at 99C coretemp all day long. At 100 C it would just yell "cool me off !" and then several deg above that it would trip. ( I have read some things that suggest another 10-20C over Tjunction but I cannot prove it). Unclewebb ran his C2D without a heatsink and made it trip/shutdown to verify a lot of what I said. Now thats some balls.

Just another note, I recommend a 65-70C working max core temp for dual cores (Tjunction 85C). Note also that coretemp and other programs average the raw data over time, usually a second. It is possbile have dangerious temp spikes that last only milliseconds. Not having a quad I dont really have a recommedation but now that you know what is going on you can make up your own mind how hard you want to push it.

But all things in moderation. (unless you have big steaming piles of money laying around doing nothing :) )
 
Thanks Billparrish,

Now I understand. I thought it was a real time reading. Now I know that it is the max temp before total shutdown.


Thanks again,

Spartus4
 
Back
Top