Sign in with
Sign up | Sign in
Your question
Solved

Tj Max , Tcase and processor degradation!

Last response: in CPUs
Share
November 13, 2012 4:06:23 PM

Trying to understand what is TjMax and Tcase , so i went on and tried chatting with intel support.


That was the text.

Quote:

Intel: Please wait for a site operator to respond. Your Guide Me session information is being transferred to the next operator.
Intel: Please wait for a site operator to respond. You are now chatting with 'Daniel'
Guest: TJ max , Tcase for 3960x
Daniel Hello. Thank you for using the Intel Customer Chat Support service. We are glad to be of service. How may I help you?
Guest: As what i have red on yout website , Tj max is only for laptops.
Guest: I have a 3960x
Daniel correct
Daniel what is your question ?
Guest: Can you explain to me the temperature limits for my processor ?
Guest: I am using Real temp 3.7
Daniel First, we don't rely on monitoring software applications from third party companies,
Guest: ok , recommend a monitoring software then
Daniel I can tell the normal operating temperature is 66.8°C and can go up to 100 degrees and eventually will shut down to prevent damage to the processor
Daniel we recommend checking the temperature directly from the BIOS
Guest: How can i check my temperature from the bios while on windows enviroment ?
Daniel What motherboard do you have ?
Guest: Rampage 4 extreme
Daniel from our side you can use the processor diagnostic tool but for directly monitoring we recommend checking with Asus also:
Daniel http://www.intel.com/support/processors/sb/CS-031726.ht...
Guest: ok.
Guest: it is download and installing at the moment
Daniel any other inquiry you may have ?
Guest: Yes ,
Guest: At what temperature i start degrading my processor ?
Daniel there is no such temperature, processors do not degrade since they are designed to work at high temperatures.
Guest: ok , then why does the processor thermal throttle at a certain temperature ?
Daniel Please provide a detail description of the issue you are experiencing.
Guest: I am not having an issue , i am trying to avoid damaging my 1000 processor.
Guest: That is why i am asking to know about temperatures.
Daniel To confirm, processors are designed to work at high temperatures so if there is any issue let us know.
Guest: Are those temperatures beyond 1000 c ?
Daniel the processor will automatically shut down when reaching 100C to prevent damage.
Guest: ok. ok.
Guest: ok , can you calrify what is the 66.8 ?
Daniel is the normal operating temperature, what clarification you need to know ?
Daniel meaning if the processor reaches this temperature or higher is considered normal.
Guest: http://www.intel.com/support/processors/sb/cs-033342.ht... http://www.intel.com/support/processors/sb/cs-033342.ht...
Guest: TCASE Max is the maximum temperature that the TCASE sensor should reach. Both TCASE and the thermal specification information ca
Guest: max should reach does not sound like "normal"
Guest: Any way , thanks for your time.
Daniel Thank you for using Intel's live chat. If you are satisfied with our support, please score items with 5 in the online survey that will be shown once you close this chat window. We value your feedback and we will use it to improve our services.



I do not know about you guys , but for me that was just a little bit hard to understand.

:) 
a c 202 à CPUs
a b å Intel
November 13, 2012 4:21:24 PM

It is quiet simple, he is saying that according to Intel the optimal temperature is 66.2degrees C, BUT thats just because he is paid to say that. LOL

Lifespan is all you can measure the limits in, as he said that above hundred degrees the CPU will shutdown to prevent damage. By adding voltages to a CPU and increasing clock you incrementally reduce the lifespan(it isn't really a degradation but more of a life). The higher you go with voltages and clockspeeds, the less lifespan a CPU will have.

Keep in mind a CPU has an effective lifespan of 10years plus, that is an age in any technological terms. By OCing a CPU you decrease that by small amount, exponentailly decreasing the higher the temps are run for extended time.

When under full stress-testing load, my suggestion is to keep it under 90 degrees C. If you can achieve that then your CPU ahs reached its max OC with your currect cooler, either get a better cooler or be happy with the OC you have.
m
0
l
November 13, 2012 7:06:12 PM

Ok.
But i sometimes read that people hate it when the temps reach 80+
Also , there is another problem.

All of those temps are what kind of temps.

Cause i have to do some sort of calibration!!!!

m
0
l
Related resources
a c 202 à CPUs
a b å Intel
November 13, 2012 8:07:09 PM

That makes no sense??? Under 80 degrees if you really want to keep it safe.. What calibration???
m
0
l
a c 139 à CPUs
a b å Intel
November 13, 2012 8:57:19 PM

Intel does a lot of research to figure out the tcase. It is recommended to not exceed this temp in a normal environment for longevity. You can't give a blanket temp for all cpus, they are all different. 80c on one cpu can cause it to die instantly while 80c on another won't even hurt it. Tjmax is usually the temp the cpu will shut off. I say usually because you can sometimes change what temp it shuts off at (this is not changing the tjmax though). Temperature is also recorded by the distance from tjmax.

If you are worried about your cpu, just don't go over tcase. Plain and simple.
m
0
l
a b à CPUs
November 14, 2012 1:19:02 AM

The TCase specification is intended for mass system builders only. That 66.8 degree number or whatever it is for your CPU is useless information for the individual end user. This temperature can only be correctly measured by cutting a groove into the top of the heat spreader and then you need to mount a calibrated thermocouple at the geometric center of your CPU. No one that just shelled out $1000 bucks for a CPU is going to take the Dremel to it No software can be used to accurately measure the TCase temperature so you are left with a number that means nothing to the individual end user.

Intel CPUs determine thermal throttling and thermal shut down based on the core temperature. The data coming from this sensor is like a reverse thermometer. As the CPU gets hotter, the data from this sensor counts down. When it reaches zero, you have reached the maximum safe operating temperature. The temperature this occurs at is commonly called TJMax or the maximum junction temperature. When it reaches this temperature, thermal throttling begins to try and keep the CPU core temperature from going any higher. Core i CPUs let you read the value for TJ Max by reading a register within the CPU. This register is fully documented by Intel for desktop and mobile CPUs. The throttling temperature is approximately 98C to 105C for the majority of their recent CPUs but you need to use software like RealTemp or Core Temp, etc. to find out what this value is.

The thermal throttling temperature has nothing to do with the thermal shutdown temperature. There is a signal in the CPU called THERMTRIP. This goes active and forces the CPU to do a thermal shutdown when the core temperature gets approximately 25C higher than the thermal throttling temperature. If the throttling temperature is 100C to 105C then the shut down temperature is not until 125C to 130C.

Asus software like their AISuite II shows more useless temperature information. The sensor being sampled may be 100% accurate but it is so far away from the hottest spot on the core that it doesn't matter what it says. It's just a number. You need to use core temperature monitoring software like RealTemp or Core Temp.

Too bad Intel doesn't hire someone to answer their phones that has actually read some of their documentation. The person you spoke to had about as much useful information about Intel CPUs as the greeter at Wally-Mart has.
m
0
l
November 14, 2012 6:36:15 AM

Let me get this straight ,
You are basically saying..
unclewebb said:
The TCase specification is intended for mass system builders only. That 66.8 degree number or whatever it is for your CPU is useless information for the individual end user. This temperature can only be correctly measured by cutting a groove into the top of the heat spreader and then you need to mount a calibrated thermocouple at the geometric center of your CPU. No one that just shelled out $1000 bucks for a CPU is going to take the Dremel to it No software can be used to accurately measure the TCase temperature so you are left with a number that means nothing to the individual end user.


Tcase is the outer temperature of the processor and it is lower than core temperature , and can't be measured in any easy way.

unclewebb said:
Intel CPUs determine thermal throttling and thermal shut down based on the core temperature. The data coming from this sensor is like a reverse thermometer. As the CPU gets hotter, the data from this sensor counts down. When it reaches zero, you have reached the maximum safe operating temperature. The temperature this occurs at is commonly called TJMax or the maximum junction temperature. When it reaches this temperature, thermal throttling begins to try and keep the CPU core temperature from going any higher. Core i CPUs let you read the value for TJ Max by reading a register within the CPU. This register is fully documented by Intel for desktop and mobile CPUs. The throttling temperature is approximately 98C to 105C for the majority of their recent CPUs but you need to use software like RealTemp or Core Temp, etc. to find out what this value is.


Tcore which has the Thermal throttle/thermal shutdown limits is measured from a sensor in the CPU cores


unclewebb said:
The thermal throttling temperature has nothing to do with the thermal shutdown temperature. There is a signal in the CPU called THERMTRIP. This goes active and forces the CPU to do a thermal shutdown when the core temperature gets approximately 25C higher than the thermal throttling temperature. If the throttling temperature is 100C to 105C then the shut down temperature is not until 125C to 130C.


Thermal shutdown limit once reached it forces CPU shutdown.
and the limit is around 125~130c which is measured through Tcore sensors.

unclewebb said:
Asus software like their AISuite II shows more useless temperature information. The sensor being sampled may be 100% accurate but it is so far away from the hottest spot on the core that it doesn't matter what it says. It's just a number. You need to use core temperature monitoring software like RealTemp or Core Temp.


When monitoring CPU temps i need to ignore useless temperatures , and i only need to concentrate on Tcore.

unclewebb said:
Too bad Intel doesn't hire someone to answer their phones that has actually read some of their documentation. The person you spoke to had about as much useful information about Intel CPUs as the greeter at Wally-Mart has.

Intel support are noobs :D 



If i got all of those information right ,

then i have few other questions/requests.
- In a monitoring application like MWMonitor there is a temperature called "package" and i assumed that this refers to Tcase which in this case as high as the highest core, am i correct and if so , how can this be higher than other core temps, if not , then what does this "package" temperature refers to?

- When setting Tj max on realtemp , the reading varies , i do not understand this , can you please further explain that?
- What is the proper Tj max temp for 3960x ?
- Can you post official links to the documentations that you are talking about that includes those information?
- Can you talk or post a link about temp-to-life spawn a voltage-to-life spawn charts for this processor?


BTW what you have said makes more sense , i used to have a laptop that can reach 100 (max readable temp) and stay on 100 (probably more) for a while untill it restarts

At last thank you for your time.
:) 
m
0
l

Best solution

a b à CPUs
November 14, 2012 3:58:35 PM

Intel realized that with multi core processors, having to monitor 6 or more cores was going to be "too much information" for many users so they decided to simplify things. All cores are sampled internally and the highest core temperature is written to a new register within the CPU so monitoring software only needs to read this single register.

The CPU Package temperature represents the peak core temperature of the CPU. Unfortunately, many users assume that the Package temperature is the same as what used to be called the CPU temperature. This CPU temperature sensor is what AI Suite II is reporting information from and is the one that should be avoided. The Package temperature shown above is what you should be paying attention to.

On my Core i5-3570K, I find that HWMonitor is reporting the individual CPU cores up to 5C or 6C too low but it usually gets the CPU Package temperature correctly or close to it. You might want to try Open Hardware Monitor too which does a better job of getting the individual cores correct if you are interested in that.

You should not touch the TJ Max adjustment in RealTemp. This feature was for the previous generation Core 2 CPUs. Core i CPUs store the correct TJ Max value in a register that any monitoring software can read. Download a fresh copy of RealTemp and let RealTemp read this information from your CPU and then don't adjust it after that.

Core Temperature = TJ Max - Digital Thermal Sensor

Intel does not use a traditional thermometer in their CPUs. It's more like a reverse thermometer which counts down as the CPU gets hotter and counts up as the CPU cools down. To convert this sensor data to the appropriate core temperature, software uses the above formula. If you go into RealTemp and start randomly adjusting the TJ Max value, you are going to screw up the reported core temperature. HWMonitor also has this "feature" but it is more hidden so users are less likely to screw things up. The HWMonitor config file has TJ Max information in it.

RealTemp GT 3.70 should read your CPU core temperatures correctly and works better than HWMonitor when the CPU is fully loaded.

I don't have access to any lifespan data for your CPU. If you operate your CPU within the Intel specs it will last a long time. Check out the forums. A lot of crazy people are doing a lot of crazy stuff to their CPUs everyday but even so, the failure rate seems tiny compared to how many "enthusiasts" are pushing their Intel CPUs to the limit of self destruction.

This is all you really need to know.

http://img838.imageshack.us/img838/5931/thermtrip.png

Processor has reached its "maximum safe operating temperature". Intel says anything below the TJ Max temperature is SAFE. In the Thermal Status area of RealTemp, it will show OK if you have not reached the throttling temperature. If you do reach the throttling temperature, even for a millisecond, the CPU throws a flag within the CPU and then RealTemp will report LOG which indicates a thermal throttling episode has been logged. If the CPU continues to throttle, you may see RealTemp report HOT which indicates that thermal throttling is presently in progress. Thermal throttling works so well that you will rarely see HOT because the CPU does a great job of keeping the core temperature on the safe side of the throttling temperature.

Here's an example of a Core 2 CPU, over volted and overclocked, happily running Prime95 with the CPU fan disconnected for 3 hours. That CPU definitely got HOT. :) 

http://img11.imageshack.us/img11/276/hote8400fw5.png
Share
November 15, 2012 2:17:18 PM

Best answer selected by Dropz.
m
0
l
!