SIO is the three-letter acronym for the Super I/O integrated circuit. For general information about this type of device, consult this page: Super I/O - Wikipedia, the free encyclopedia.
On the 8 Series boards, an eSIO device is utilized, the Nuvoton NCT6683D. An eSIO device is an SIO device that contains an embedded microprocessor (known as a dedicated microcontroller). This architecture allows both traditional and nontraditional SIO functions to be fully or partially implemented in firmware. This allows more sophisticated capabilities to be (cost-effectively) implemented than would be possible using only discrete circuitry. In the NCT6683D parts utilized on the 8 Series boards, Intel worked with Nuvoton to implement a custom firmware package for the device that allows Intel Quiet System Technology (Intel® QST) to be utilized. Intel® QST provides sophisticated fan speed control algorithms that can dramatically improve overall system acoustics and psychoacoustics while still ensuring proper temperatures. To better understand psychoacoustics, consult this page: Psychoacoustics - Wikipedia, the free encyclopedia.
Because the eSIO device contains a microcontroller, it can run hotter than traditional SIO devices. As a result, a temperature sensor was included in the device so that the temperature of the eSIO device itself can be monitored and managed by the Intel® QST algorithms. Because a heatsink is not typically installed on this device, its idle temperature is a little higher than most other motherboard devices. A reading of 48°C is perfectly normal, as is a reading of 65°C or even 70°C in stressful conditions. These higher readings in stressful conditions are actually the result of heat bleeding from other devices near the eSIO and increasing its temperature. To put it all in perspective, a critical temperature for this device is way up at 95°C. By default, the Intel® QST algorithms are configured to keep this device at or below 85°C.
Thanks Mr. Scott for the detailed explanation and assurance that SIO temp is normal and nothing to panic.
I have also noticed that the RAM voltage fluctuates between 1.480 and 1.510 volts (every 3 seconds) on this DH87RL - that's a 0.030 volt variance.
Is this also normal?. The ram is configured with XMP profile and the corsair vengeance ram has a full compatibility with i3/i5/i7 processors.
I have DH77DL, DH67CL, DH77EB boards and all of them report a constant 1.50 volts. I am using the same corsair vengeance ram (2 x 8GB Sticks) 1600Mhz on all of the systems. They are all working perfectly for the last 3 years.
Yes, this is perfectly normal and expected.
The AtoD (Analog to Digital) converter used in the NCT6683D eSIO supports a limited voltage range. For most voltages, this means that the least-significant bit (LSB) of a reading represents 0.016v. It would thus not be unexpected to see variances in readings of +/- 0.016v as the voltage hovers between adjacent readings. Some voltages must be scaled down (using a resistor pair) to fit within the limited voltage range of the AtoD and then, when readings are taken, the readings will need to be scaled back up to the proper voltage range. For the 12.0v sense, a scaling factor (multiplier) of 12 is required. This means that the LSB represents 0.192v. For the 5.0v sense, a scaling factor of 5.02 is required. This means that the LSB represents 0.08032v. For the CPU Input and Memory voltages, a scaling factor of 1.5 is required. This means that the LSB represents 0.024v.
Please note that these scaling factors are specific to the NCT6683D eSIO device and the resistor network used on the 8 Series boards. On 7 Series and older boards, the scaling factors are different; specific to the particular SIO device and the resistor network used on the board in question.
P.S. To be complete, for those folks that have DZ87KLT-75K boards, there are an additional 9 CPU voltages monitored. For the CPU Core 0-3, CPU Graphics and CPU Ring voltages, a scaling factor of 1.5 is required. The CPU System Agent, CPU I/O and CPU PLL voltages need no scaling factor.
I thought that this was worth also mentioning...
As you might imagine, most of the general-purpose monitoring applications (CPUID's HW Monitor and PC Wizard applications, Open HW Monitor, Hmonitor, HWiNFO32/64 and many others) have problems accurately displaying voltage sensor readings on many boards (not just Intel's). This is due to the fact that they do not know (a) what sensors in the SIO/eSIO are used for what voltages and (b) what the resistor network is that is utilized on the board. Before I retired, one of the things that I did at Intel was to attempt to communicate this information to these software vendors. I was successful in getting a couple of these software vendors to build tables into their applications so that they can get this right. The best example is the AIDA64 application, which perfectly handles all of the Intel Desktop Boards and those NUCs that were released before I retired (and hopefully those after if my former co-workers have kept up this process). It is thus my top recommended application. I also worked with Alfredo Milani-Comparetti, the author of the SpeedFan application. At the time of my retirement, he was very close to having all of the Intel Desktop Boards handled, as well as some of the NUCs. Speedfan is thus my second-most recommended application.
P.S. I am probably going to get a bunch of nastigrams from some of the authors for these other applications. If they have handled this issue, sorry for not pointing it out accurately; I don't spend a lot of time regularly trying out any new versions of these applications...