Intel Temperature Guide - by
CompuTronix
Rev. 30912
Preface:
The purpose of this Guide is to provide Intel overclocking enthusiasts with an understanding of specifications, thermal relationships and standard test methods, so that temperatures can be properly tested and analyzed. This Guide supports Core i5 "K" series and Core i7 "K" series 2nd Generation Sand Bridge, 3rd Generation Ivy Bridge and 4th Generation Haswell desktop processors.
Introduction:
Intel's Thermal Specification - http://ark.intel.com/ - for each processor variant is shown as "Tcase". Although this specification may seem simple enough, the technical details are very complicated, so temperature issues continue to be a major source of confusion and debate in the overclocking enthusiast community.
In order to gain a clear perspective, it's important to keep the terminology straight. There are 5 thermal sensors in a 4 core processor; a single Analog sensor (CPU temperature) and 4 individual Digital sensors (Core temperatures).
CPU Temperature:
The temperature shown in Intel's specification, (Tcase), is measured on the surface of the Integrated Heat Spreader (IHS) under tightly controlled laboratory conditions at 22C Standard Ambient (intake temperature). For lab testing
only, a groove is cut into the surface of the IHS where a thermocouple is embedded at the center. The stock cooler is attached, and a steady-state 100% workload is applied, (such as Prime95 Small FFT's). Thermal saturation is reached within 10 minutes.
Since there is no thermocouple on any processors outside Intel's labs in the wholesale or retail outlets, a single Analog Thermal Diode is instead used to "emulate" a thermocouple. This single Analog sensor is located in the center of the lower layers of the processor package and is called "CPU" temperature, which is the equivalent of "Tcase".
The Analog value is converted to Digital (A to D) by the Super I/O (Input / Output) chip on the motherboard, then is calibrated to look-up tables coded into BIOS,
which are all too often inaccurate. This is the temperature you see in BIOS and in monitoring utilities such as AI Suite II or OC Panel, which are provided by motherboard manufacturers.
Core Temperature:
Also called "Tjunction", Core temperature is measured at the heat source by individual Digital Thermal Sensors (DTS) for each Core, which are factory calibrated by Intel. Since the CPU sensor is not in close proximity to the heat sources, there is ~5C thermal gradient or "offset" between "Core" temperature and "CPU" temperature during Prime95 Small FFT's.
This means that "CPU" temperature runs ~5C lower than "Core" temperatures. More correctly, since there is always some temperature sensor discrepancies between Cores, "average" Core temperature is more important.
To simplify Intel's Thermal Specification, at Default / Auto BIOS settings (stock clock and Vcore) with the stock cooler, if Prime95 Small FFT's is run for 10 minutes at 22C Ambient, then the average "Core" temperature would be ~5C higher than "CPU" temperature.
Examples:
2nd Generation i5 2500K / i7 2700K (95 Watts)
Standard Ambient = 22C
Tcase (CPU Temp) =
72C
CPU / Core Offset + 5C
Tjunction (Core Temp) =
77C
Tj Max (Shutdown Temp) =
98C
3rd Generation i5 3570K / i7 3770K (77 Watts)
Standard Ambient = 22C
Tcase (CPU Temp) =
67C
CPU / Core Offset + 5C
Tjunction (Core Temp) =
72C
Tj Max (Shutdown Temp) =
105C
4th Generation i5 4670K / i7 4770K (84 Watts)
Standard Ambient = 22C
Tcase (CPU Temp) =
72C
CPU / Core Offset + 5C
Tjunction (Core Temp) =
77C
Tj Max (Shutdown Temp) =
105C
Shutdown Temperature:
Tj Max is the Thermal Specification which defines the "Core" temperature at which the processor will shutdown. Although Intel's Core i processors are capable of operating above 90C, we also know that excessive heat kills electronics. As common sense and experience usually prevail, most folks in the Intel overclocking community will agree that
sustained Core temperatures greater than high 70's are too hot for stable gaming, a 12 hour stability test, or for processor longevity.
Ambient Temperature:
"Ambient" temperature means "room" temperature, or more correctly, "intake" temperature at or near your computer's air intake. This is a very important temperature measurement, because
Ambient directly affects all other computer temperatures. Use a trusted analog or digital indoor thermometer to measure Ambient temperature. When performing a Thermal Test, if you're testing above 22C Standard Ambient, then you need to subtract the difference so that your test results are corrected to Intel's standard. This way you eliminate variables, and your results will always be consistent and repeatable. Here's the conversions and a short scale:
F-32/9x5=C ... or ... Cx9/5+32=F ... or more simply ... 1C = 1.8F
20.0C = 68.0F (Cool)
21.0C = 69.8F
22.0C = 71.6F (Standard) ... or ... 22.2C = 72.0F
23.0C = 73.4F
24.0C = 75.2F
25.0C = 77.0F (Warm)
Overclocking and Vcore
Power dissipation can reach 130% to 150% of your processor's TDP envelope when Overclocked with either Auto or Manual Vcore settings, so high-end cooling solutions are required. Regardless, excessive Vcore and temperatures will result in accelerated "Electromigration", which prematurely erodes the Traces and Junctions within the processor's layers and nano-circuits. This will in turn eventually result in Blue-Screen crashes, which will become increasingly frequent over time. Therefore, Vcore settings should not exceed the following:
45 Nanometer 1st Generation Core i ..... 1.40 Vcore
32 Nanometer 2nd Generation Core i .... 1.35 Vcore
22 Nanometer 3rd Generation Core i ..... 1.30 Vcore
22 Nanometer 4th Generation Core i ..... 1.30 Vcore
3rd and 4th Generation Core i processors use a poor quality, unevenly spread Thermal Compound between the top of the Cores and the underside of the Integrated Heat Spreader (IHS). Consequently, some processors show a wide variation between Core temperatures, or one Core which runs significantly hotter than it's neighbors.
This has prompted some enthusiasts to "de-lid" or remove their processor's IHS to replace Intel's Thermal Compound with a high quality Thermal Compound. However, 2nd Generation Core i and earlier processors don't suffer from these problems, as Intel used fluxless solder, which provides efficient heat transfer.
Thermal Testing @ 100% Workload:
Prime95
Small FFT's is the standard for processor Thermal Testing, because it's a steady-state 100% workload. This is the test that Real Temp uses to test sensor movement. Prime95's default test,
Blend, is a cyclic workload for testing memory stability, and is
not suitable for thermal testing. Other tests such as LINPACK and IBT (Intel Burn Test is NOT written by Intel) have segments that load all registers with all one's, which is equivalent to a 115% workload. While suitable for stability testing, IBT is not a steady-state 100% workload.
You can see for yourself by using the "Graphs" in SpeedFan to illustrate and compare how these tests create different thermal signatures. On air cooling, thermal saturation is typically reached within 10 minutes, however on liquid cooling, up to 30 minutes may be required to reach saturation. Testing should be conducted with the computer clear of any desk enclosures, case covers open and all fans (and if equipped with liquid cooling, the circulating pump) at 100% so that base-line temperatures can be seen and noted as the best-case scenario.
Thermal Testing @ Idle:
If "Speedstep" is "Disabled" in BIOS, then depending upon Vcore and clock speed, Idle Power can be nearly 40 Watts, which will result in high Idle temperatures, especially when combined with a high Ambient temperature. Speedstep needs to be Enabled to achieve the lowest possible Idle temperatures. Idle means
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.
Allow your rig to "settle" for 10 minutes on air cooling and up to 30 minutes on liquid cooling, then measure Ambient temperature. Use CPU-Z to confirm that Speedstep has dropped the frequency to 1600 MHz, and use Core Temp, Real Temp or Hardware Monitor to confirm that Power is ~ 6 Watts. Under these controlled conditions, an accurate Core temperature will indicate ~ 6 to 7C above Ambient.
Core Temperature sensors are designed to be more accurate at high temperatures for shutdown protection (TjMax). Many 45, 32 and 22 Nanometer processor variants have sensors with large "slope errors" or sensors that "stick" at 10c or more above Idle. Sensor movement can be tested using Real Temp. It's common to see a Core or two that won't indicate a true Idle temperature.
__________________________________________________
Thank you for reading. I hope this Guide answers any questions you may have about processor temperatures.
CompuTronix
I'm delidded and running naked on a high end water setup so my temps are great.
Facebook
Twitter
Instagram