Intel Temperature Guide -
Continued
Section 9 - The 22 Nanometer Problem
Core i 3rd and 4th Generation processors are very sensitive to small increases in voltage and frequency. When overclocked, temperatures could rise above 90C, so
high-end air or liquid cooling is critical. 22 Nanometer processors are more difficult to cool than previous Generations for three reasons:
(1) The 3rd and 4th Generation 22 Nanometer Die has 35% less surface area in contact with the underside of the Integrated Heat Spreader (IHS) than the 2nd Generation 32 Nanometer Die.
(2) 3rd and 4th Generation processors have over 20% more transistors packed into a smaller Die than 2nd Generation processors.
(3) 3rd and 4th Generation processors use Thermal Interface Material (TIM) between the top of the Die and the underside of the IHS instead of solder, which was used in 2nd Generation and earlier processors.
![]()
Since the bonding material which seals the perimeter of the IHS to the Substrate is slightly too thick, this tends to increase the space between the underside of the IHS and the Die, which can cause the TIM to compress unevenly. The effect of this manufacturing procedure is that many processors show a wide deviation between Core temperatures, or one Core which runs much hotter than it's neighbors.
This has encouraged some overclockers to "de-lid" or remove their processor's IHS, which basically involves thoroughly removing the bonding material, replacing only the TIM and then restoring the IHS. Typical results are significantly lower Core temperatures and less deviation between Cores. Here's an excellent YouTube -
http://www.youtube.com/watch?v=XXs0I5kuoX4 - that shows before, how-to, and after, but beware that de-lidding will void your warranty, and it's not difficult to damage or destroy your processor.
Intel has addressed these thermal problems in their Haswell refresh. The Devil's Canyon processors have an improved IHS alloy and a new Polymer TIM. Although not as thermally efficient as solder, temperatures have been improved by several degrees.
Regardless, 4th Generation processors differ from their 3rd Generation counterparts in that they have a Fully Integrated Voltage Regulator (FIVR) on the Die, instead of on the motherboard. This increases their Thermal Design Power(TDP), so all other factors being equal,
4th Generation processors run hotter at 100% workload than any of their predecessors.
Section 10 - Software Tools
In order to properly test and evaluate your temperatures whether overclocked or not, you'll need to download the following freeware utilities:
CPU-Z -
http://www.cpuid.com/softwares/cpu-z.html
Hardware Monitor -
http://www.cpuid.com/softwares/hwmonitor.html
Prime95 v26.6 -
http://windows-downloads-center.blogspot.com/2011/04/pr...
Real Temp -
http://www.techpowerup.com/downloads/2089/real-temp-3-7...
SpeedFan -
http://www.almico.com/sfdownload.php
Section 11 - Thermal Testing @ 100% Workload
We all remember science class where one of the basic principals for conducting any scientifically controlled experiment, is that it's critical to follow the same procedure every time. This eliminated variables so results will be consistent and repeatable. If everyone is testing their rigs with X stress software at Y Ambient temperature with Z measuring software, then it's impossible to compare apples to apples.
This is why processor temperatures continue to be a major source of confusion and debate. In this Section we'll explain how to properly test your rig. It's important to remember that
we are not stress testing; we are thermal testing to produce a Core temperature benchmark.
Prime95 Small FFT's is the standard for CPU thermal testing, because it's a steady-state 100% workload. This is the test that Real Temp uses to test sensors. The link above is to version 26.6, which is well suited to all Core 2 and Core i variants.
Core i 2nd, 3rd and 4th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95 run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces unrealistically extreme temperatures. The FPU test in the software utility AIDA64 shows the same results.
It's not necessary to run AVX code for thermal testing. Prime95 v26.6 produces temperatures on 3rd and 4th Generation processors more consistent with 2nd Generation, which also have AVX instructions, but do not suffer from thermal extremes due to having a soldered Integrated Heat Spreader and a 35% larger Die.
Prime95's default test, Blend, is a cyclic workload for testing memory stability, and Large FFT's combines CPU and memory tests. As such, Blend and Large FFT's both have cyclic workloads which are unsuitable for CPU thermal testing.
Other stability tests such as Linpack and Intel Burn Test have cycles that load all registers with all one's, which is equivalent to a 110% workload, and are also unsuitable for CPU thermal testing. The software utility OCCT runs elements of Linpack and Prime95.
![]()
Shown above from left to right: Small FFT's, Blend, Linpack and Intel Burn Test.
![]()
Shown above from left to right: Small FFT's, Intel Extreme Tuning Utility CPU Test, and AIDA64 CPU Test.
The "Charts" in SpeedFan show how these tests create different thermal signatures. Intel Extreme Tuning Utility is also a cyclic workload. Although AIDA64's CPU test is steady-state, the workload is insufficient, and it's not available as freeware.
Setup:
Testing should be performed with your computer clear of desk enclosures or items that block airflow. Covers should be removed and all fans and circulating pump (if equipped with liquid cooling) at 100% RPM, so temperatures can be tested under ideal conditions.
Testing close to 22C Ambient is preferred so as to provide normal thermal headroom, but is not required. Testing at high Ambient temperatures should be avoided. If adequate A/C is unavailable, then test late at night or early in the morning when Ambient is lowest.
When performing a thermal test above 22C Ambient, remember to subtract the difference so that test results are corrected to Standard Ambient. This eliminates variables so results will be consistent and repeatable.
Example:
During Thermal Testing @ 100% Workload;
if
measured Ambient is 25C, and
reported Core temperature is 80C,
then at
standard Ambient 22C,
corrected Core temperature would be 77C.
Test:
Run Prime95 v26.6 Small FFT's for 10 minutes, then use your thermometer to measure Ambient. Use Real Temp to measure your Core temperatures. Correct your results to Standard Ambient.
Results:
Core temperatures
should be in the mid 70's for most processor variants. However, depending on variables such as clock speed, Vcore, Ambient and cooling, Core i 3rd Generation, and especially 4th Generation temperatures could rise above 90C. Core temperatures in the
mid 70's are safe during real-world everyday workloads.
Exception: Early Core 2 1st Generation processors with stepping revisions B2 and B3 and have lower Tcase and TjMax specifications than other variants. Core temperatures on these processors should be kept under 70C.
Deviations in sensor accuracy between individual Cores can be up to +/- 5C. This means there could be 10C between the highest and lowest Cores, so "average" Core temperature is often more realistic.
Section 12 - Thermal Testing @ Idle
Look closely at the SpeedFan Charts above, where idle temperatures are shown between load temperatures. Notice that some Cores have more "range" than others and idle lower. Sensors can be tested with Real Temp. Core temperature sensors are designed to be more accurate at high temperatures for Throttle protection, so
don't expect idle temperatures to be accurate.
Remember that when you power up your rig from a cold start, all components are at Ambient, so temperatures can only go up. With conventional air or liquid cooling,
no temperatures can be less than or equal to Ambient.
If "Speedstep", also called Enhanced Intel Speedstep Technology (EIST), is disabled in BIOS, then depending on Vcore and clock speed, idle Power can be nearly 40 Watts, which will result in high idle temperatures, especially when combined with high Ambient temperature.
Setup:
Speedstep and all "C" States (C1E on earlier motherboards) needs to be enabled to achieve the lowest possible idle temperatures. Also, if Windows Power Options is not set correctly, then Speedstep will not work.
To check this, click on Control Panel, Power Options, then to the right of the selected plan, click on Change plan setting. Next click on Change advanced power settings, then drag the scroll bar down. Click on + next to Processor power management, then click on + next to Minimum processor state. This Setting must be 5%. If it's not, then correct it and click Apply.
Restart into BIOS and confirm that you've saved your settings to a Profile. Change all settings to stock (Default / Auto) including SpeedStep, all C States and Vcore, then save and exit. Reboot into Windows and confirm that your rig is at
dead idle; no programs running, and off line. No Folding or SETI or tray trash running in the background, and less than 2% CPU Usage under the "Performance" tab in Windows Task Manager.
Use CPU-Z to confirm that Core Voltage and Core Speed has decreased as follows:
-> Core 2
1st. Generation 65 Nanometer ... less than 1.250 Volts @ 1600 Mhz
2nd Generation 45 Nanometer ... less than 1.100 Volts @ 2000 Mhz
-> Core i
1st. Generation 45 Nanometer ... less than 1.000 Volts @ 1600 Mhz
2nd Generation 32 Nanometer ... less than 1.000 Volts @ 1600 MHz
3rd Generation 22 Nanometer ... less than 0.900 Volts @ 1600 MHz
4th Generation 22 Nanometer ... less than 0.800 Volts @ 800 MHz
Use Hardware Monitor to confirm that Power has decreased as follows:
-> Core i
2nd Generation 32 Nanometer ... less than 8 Watts
3rd Generation 22 Nanometer ... less than 4 Watts
4th Generation 22 Nanometer ... less than 2 Watts
Test:
Allow your rig to "settle" for 10 minutes, then measure Ambient temperature. Use Real Temp to measure your Core temperatures. Remember to correct your results to Standard Ambient. Power (watts) isn't monitored on Core 2 processors, but for general reference, idle power for several popular CPU's is shown above under Section 7 - Relative Temperatures.
Results:
At stock settings, Core i 2nd, 3rd and 4th Generation processors
should idle at less than 8C above Ambient. This means that at 22C Standard Ambient your Cores
should idle just under 30C. Certain Core 2 variants and Core i 1st Generation variants may idle a few degrees higher.
The better your cooler and the lower your idle power, the lower your idle temperatures.
Section 13 - Improving Temperatures
Whether your computer is a stock workstation or an overclocked gaming rig, achieving the lowest possible temperatures always depends on components, configuration and air flow. Here's a few thoughts:
* Intel's cooler is barely adequate at stock. If you intend to overclock then upgrade your cooler.
* Good cable management yields good airflow. Use zip-ties, patience and attention to detail.
* Axial flow graphics cards recirculate heat. Linear flow cards exhaust heat from the case.
Examples:
Axial -
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
Linear -
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
* Axial cards work well with a liquid cooled CPU. Linear cards work well with an air cooled CPU.
* A hot case stresses hard drives, memory, chipsets, voltage regulators and power supply.
* If your case doesn't breath well, then perhaps it's time to upgrade to one that does.
* High performance computers need unrestricted airflow in and out, so location is critical.
* Good fans are important, but if you want a quiet computer, then consider a fan controller.
* If your rig runs 24/7, then hard drive and fan bearings are wearing, and dust is accumulating.
* Clean the dust out of your rig. Perform regular Planned Maintenance Inspections (PM's).
Concerning Thermal Interface Material (TIM), here's some helpful links:
Thermal Paste Comparison, Part One: Applying Grease And More -
http://www.tomshardware.com/reviews/thermal-paste-heat-...
Thermal Paste Comparison, Part Two: 39 Products Get Tested -
http://www.tomshardware.com/reviews/thermal-paste-perfo...
Section 14 - Summary
* Standard Ambient temperature is 22C.
* Ambient affects all computer temperatures.
* As Ambient increases, thermal headroom decreases.
* Don't expect BIOS or CPU temperature to be accurate.
* Core temperatures respond instantly to changes in load.
* 80C
sustained Core temperature is hot.
* Core temperatures in the mid 70's are safe.
* Excessive Vcore and temperatures accelerate electromigration.
* Prime95 Small FFT's is the standard for thermal testing.
* Deviations between individual Cores can be up to +/- 5C.
* Core temperature sensors are more accurate at high temperatures.
* Don't expect idle temperatures to be accurate.
* Sensors can be tested with Real Temp.
* No temperatures can be less than or equal to Ambient.
Section 15 - Intel Processor Temperature FAQ http://www.intel.com/support/processors/sb/CS-033342.ht...
_________________________________________________
Thank you for reading. I hope this Guide has answered your temperature questions.
CompuTronix