Definitive answer:85c on C2d with TAT?? - Page 2 - Overclock.net

kevsta112003 · 10-16-06

Still confused.

Intel took thermal specification from previous generations of chips (i.e. Prescotts/Preslers) using the temp recorded from the geometric top of the IHS. For example, this p4 ( http://processorfinder.intel.com/det...x?sSpec=SL8J7# ) has a 11c difference in thermal specification than a core 2 duo. However, a stock Core 2 Duo at load (measured using speedfan) is significantly cooler than a stock pentium 4 at load (measured by speedfan). People used speedfan to determine safe temps for past generations of processors, regardless of the location of where the thermal specification was taken from. If the thermal specification for the new Core 2 Duos is taken from the same spot as previous generations of processors, and is 10c lower.

I think a good test would be to compare the temp taken from the geometrice center of the IHS (as you did) from a previous generation of intel CPUS, and compare it to temps recorded in speedfan.

Thumper · 10-16-06

People took the safe temps measured from wherever they could previous. Short of installing a thermal probe permanently, there was no way to read temps other than through the sensor provided by the motherboard. All Intel specifications for temps have been given as being read from the center of the IHS, regardless of where we could read the temp from. In fact, I would not be suprised to see a larger variance in the difference between the IHS and speedfan in previous CPUs simply because those temps are read from the motherboard. Again, this doesn't mean that considering the speedfan temp as gospel in previous generations was right.

Just because the Core 2 Duo runs cooler at stock than the previous CPUs does not mean it can go as hot or hotter than they could. First off, you can't compare the Allendale and Conroe CPUs to a Prescott, one is 65nm the other is 90nm.

Also, the Core 2 Duo runs cooler due to lower voltage and better thermal control, again, this does not mean that they can go hotter than before.

I would be willing to run similar tests on my Pentium D 920 and my Pentium 4 Northwood, however I really can't do it soon as I am under the wire in the case mod contest and I would like to find the baseline voltage on my 6700 and get to overclocking it, plus I am doing a case for a buddy now too.

My primary concern was getting some solid information on the way the temps are calculated on these new CPUs as without it, there would have been many dead chips in a short while after being cooked 24/7 at 70c+.

455buick · 10-16-06

Quote:

Originally Posted by Thumper

The question.

Is it safe to run a Core 2 Duo at up to 85c as measured by the Thermal Analysis Tool from Intel?

To answer this question, we need to determine if the temp of the core (which is measured by TAT) is different than the temp of the center of the IHS, as this is where Intel gives us the max safe threshold.

I got everything set up and ran some quick tests. Unfortunatly the sensor wire kept the heatsink from sitting flush so I could not run a load test, but did take measurements with idle and 50% on one core. I tried placing the sensor 2 different ways neither had any impact on temps or seating of the heatsink.

Here are the results;

This is proof of the placement of a thermal sensor on my IHS, positioned to read the center.

Thermal Probe Placement

Measurement of the core through TAT at idle in the destop:

Core Temp Idle

As you can see,Core 1 reads 51c and Core 2 reads 45c. I beleive the difference in temps is due to the seating of the heatsink, without it the two cores are usually no more than 3c off each other.

Now, here is what the thermal probe reads at the same time as the screenshot was taken.

Thermal Probe Idle

It varied from 51-54c over a minute long period.

Here is the TAT reading with 1 Core at 100% (50% CPU load)

Core Temp 50%

It clearly shows Core 1 at 57c and Core 2 at 54c. These readings bounced slightly going as high as 60c/56c.

Now, here is the thermal probe reading for the same time.

Thermal Probe 50%

It clearly shows 59c, it also varied slightly over a minute period, going as high as 63c.

Personally, this is fairly conclusive evidence that there is very little difference between the temps reported by TAT and those seen at the center of the IHS. In fact there was very little difference between the temps reported by TAT, and by PC Probe, within 3c of each other usually.

Since the recomended guidelines for a Core 2 Duo stated by Intel, and measured at the center of the IHS are approx 60c, and the temps reported in the cores are almost identical to those found on the IHS center, there is no support for safe operation of a Core 2 Duo over 60c, regardless of the program used to measure the temp.

It should also be noted, that the 60c mark is listed by Intel for a CPU being operated at stock settings. For an oveclocked Core 2 Duo, within Intel voltage specs, 55c is the maximum I would recomend for 24/7 operation.

Excellent post Thumper!

This should help everyone with regards to core 2 temps. I suspect that people searching for this data on the internet, will most likely be brought here to overclock.net. I hope they find your data useful as well.

As in all cases regarding cpu temps, and gpu temps (any cpu or gpu!) the manufacturer's recommendations should be followed and not over looked. Unless you like buying parts and pieces for your computer. Especially those highly expensive ones. ;-)

Take care, :-)

pauldovi · 10-18-06

They have Kentsfields doing 90-100C. I am sure a Conroe can handle 85C. Most of your Dells and notebooks are sitting well above 60C idle. My dad's IBM T60p with a T2700 in it does 63C idle right after startup. My friend's MacBook Pro does 71C idle with a T2600.

Thumper · 10-18-06

#1 the "T" series Core 2's are mobile CPUs, designed differently, and have differnet thermal maximums listed by Intel. I doubt Intel is going to state such different levels for identical CPUs......

#2 The point of my tests was not to see what they could "handle", my point was to show what was safe.

A car can "handle" driving 50,000 miles without ever changing the oil.......but you better hope you aren't the owner of that car when it passes 55,000 miles.

If someone wants to run their $500+ CPU at 85c because it doesn't smoke instantly, that is their choice, and I will likely feel a little sorry for them when the CPU dies after a couple years (chances are they are ignoring the voltage guidelines too, so maybe only several months)........

ThaWaxShop · 10-18-06

Well first off ive done my own tests and at 80c the chip goes into throttle warning and at 85c it throttles. The sensor from TAT to pc prboe is closer because of bios updates to calibrate it or to actually read from the digital sensor

Sunburnti · 10-18-06

Well could someone please explain to me about how much on average will a CPU temp rise from "idle" to full load? I dont understand how I cna check the load temp.

When I boot my PC, I use a program I downloaded from guru3d that supposedly was made for core2duo. It always says I am arround 40-45 c I assume this is idle. Does that mean when I'm playing games I should be safe? about how much does the temp rise?

*edit. I'm running at 2.8ish gigs on my cpu, at Auto voltage.

ThaWaxShop · 10-18-06

Never use auto voltage.... its probably got you voltage to 1.6 or something. Auto sets your voltage to what the BIOS thinks it will need. Its always wrong lol

swayne · 10-18-06

i have got up to 70 load and i havent throtled at all

**The_Manual** · 10-19-06

Please be aware that the Throttle system will not become active until the processor believes that the thermal status of the chip is far outside of design.

A throttle point in reality means that the chip would be taking damage if TM1 was not in an active mode. Chips can go higher than their recommended design specifications without forcing a Thermal Control system design. This was how the processor algorithms were designed to accommodate, without letting serious damage to the transistors and processing logic cores occur.

Quote:

Well could someone please explain to me about how much on average will a CPU temp rise from "idle" to full load? I dont understand how I cna check the load temp.

You can never exactly determine the increase in thermal load from a processor idle state.
You can however determine that actual load/idle temperature of your processor within a given range without actually running the processor at a load/idle state. To determine the full load/idle state of your computer you can use this equation for thermal status equated with thermal chip resistance.

[Power (W) X C/W (Chip/Cooling Thermal Resistance)] + Ambient Temperature = Processor Temperature

You can use this equation to determine both the idle and load state temperatures.
You are then able to compare both to determine a percentage increase of thermal status to the chip in question.

An example can be provided:

100W x 0.18C/W + 28C = 46C Load
65W x 0.18C/W + 28C = 39.7C Idle

Percentage increase = 15.87% Increase in temperature Idle > Load.

In reality both the Ambient temperature and the thermal resistance will slightly change when in an idle or load state than those I have given within my example.

Quote:

They have Kentfield’s doing 90-100C. I am sure a Conroe can handle 85C. Most of your Dells and notebooks are sitting well above 60C idle. My dad's IBM T60p with a T2700 in it does 63C idle right after startup. My friend's MacBook Pro does 71C idle with a T2600.

Kentsfield is operating at 100C because there are four cores integrated in one chip that are generating thermal temperature.
Now logically this should mean that the thermal status of Conroe should double therefore reaching 120C.

However 120C is too hot for the processor to manage and therefore an increase to thermal shielding was applied to this processor to reduce the total temperature.
Due to the integration of four cores the components used to create this chip were slightly altered to withstand higher temperatures. However 100C is still in my opinion too hot even though the chip is designed to withstand that level of thermal temperature. A change to the chips thermal resistance played a large part in the additional ability of the chip to withstand temperature.

The Thermal Defence algorithms of this chip have been modified for a design of a 100C thermal design structure. Therefore safety mechanisms will be offline until around 115C. Approximately 130C will prompt all remaining auxiliary thermal defence procedures to become active to save the chip's life expectancy.

However like Intel Core 2 Duo the Core 2 Quadro has a design structure to actually withstand 1.8v for processor voltage (vcore). The transistors and onboard voltage regulation system are designed to operate with 1.8v with only a small decrease to life expectancy.
Therefore regardless of what normal overclockers say engineers will acknowledge, like myself, that 1.8v is an acceptable voltage is using advanced cooling i.e. Liquid Nitrogen.

Quote:

Intel took thermal specification from previous generations of chips (i.e. Prescotts/Preslers) using the temp recorded from the geometric top of the IHS. For example, this p4 ( http://processorfinder.intel.com/det...x?sSpec=SL8J7# ) has a 11c difference in thermal specification than a core 2 duo. However, a stock Core 2 Duo at load (measured using speedfan) is significantly cooler than a stock pentium 4 at load (measured by speedfan). People used speedfan to determine safe temps for past generations of processors, regardless of the location of where the thermal specification was taken from. If the thermal specification for the new Core 2 Duos is taken from the same spot as previous generations of processors, and is 10c lower.

I think a good test would be to compare the temp taken from the geometrice center of the IHS (as you did) from a previous generation of intel CPUS, and compare it to temps recorded in speedfan.

Please be aware that the sensor used to determine processor temperature is not the same on Intel Core 2 Duo processors as it was on Intel Pentium 4 processors.

All software that is used to determine processor thermal status will never be accurate due to the design of the algorithms and the ability of the x86 architecture to designate status to software algorithms. With VAX we would not be obtaining problems, however VAX is not used (even though its a far more powerful architecture, even though its 10+ years old).

The "significantly cooler part" is far from true. My Pentium 4 540 operated at a temperature of 60C in full load (core). Now on Core 2 Duo the sensors report 44C (board), however with TAT they report 56C at stock settings on the core. This therefore states that both processors will operate within the same thermal status, however due to the decrease in process size the thermal design is different, and also as there are two cores versus one (90nm > 65)

The thermal specification given from the Intel Processor Finder site relates to the Core temperature, regardless of the default sensor used. On Pentium 4 processors the core sensor was the default sensor. On Core 2 Duo it is not. This does not depend on design but due to the i975x, 965 etc chipset design for thermal status reporting with Core 2 Duo

		Overclock.net - Overclocking.net > Intel > Intel CPUs
Definitive answer:85c on C2d with TAT??

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)