So I've got an Intel DH87RL motherboard with an i5-4570S CPU that's been working really well for.. well, it's been a while. Anyway, I added a new GPU to it, and figured I should see what the temperatures are like on everything when it's under a full load.. and found that the CPU is hitting close to 80c without the fans ramping up AT ALL.
I tried playing around with the fan settings in the BIOS, and I found that if I switch it over to a manual fan control it does change the fan speed, so the fan control is definitely working. If I adjust the minimum duty cycle it will also increase, however it will stay at that minimum speed. The CPU fan is a 4pin, and the rear/front (both acting as intake) are both 3pin, though it seems like the fan control should still work in this case (I know the CPU requires a 4pin).
Is there something obvious that I'm missing with regard to a target temperature or something on the CPU? I'd consider replacing the intake fans with 4pin units, but even the CPU fan isn't changing speed at all, so that doesn't seem to be the fix here. I've got the minimum duty cranked up for now just to keep things reasonably cool, but I'd like to let the fan control do it's thing and ACTUALLY control the speed of the fans, based on temperature.
I am sure that there are lots of things of which you are not aware, but let's discuss only those that are pertinent to the topic at hand
The most important thing to understand is that the fan speed control solution on this board utilizes Intel's patented Quiet System Technology (QST) fan speed control algorithms. Amongst other things, these algorithms support the optimization/minimization of the acoustic penalties that can be incurred dealing with a system's thermal load and can eliminate or greatly reduce many of the psycho-acoustic effects that can occur from transients in that thermal load. Its attributes include:
Secondly, within Desktop processors, there is an additional temperature threshold, known as the Control Temperature (or Tcontrol), that has been defined. This threshold, which is determined during the thermal validation of the processor die (and is thus unique to each individual processor), defines the temperature at or below which the processor can operate constantly without suffering any thermal degradation. That is, even if the temperature was held constantly at this threshold for the processor's entire warranted lifetime, it would not suffer an thermals-related silicon degradation. It should be noted that, by definition, if processor temperatures go above the Tcontrol threshold and a full knowledge of the processor's thermal load line is not available (and it is not in most desktop system designs), then the fan speed control subsystem is required to respond aggressively - even to the point of running the processor fan at full speed - to bring the temperature back below the Tcontrol threshold.
The existence of a Tcontrol threshold melds beautifully with the use of a PID algorithm and allows us to optimize the fan response to processor temperatures. As the temperature rises closer and closer to the limit (Tcontrol), the PID algorithm will analyze the rate at which the temperature is increasing and respond in anticipation of it exceeding the limit. The level of anticipation will be used, in league with a responsiveness coefficient, to determine the magnitude of the fan response. If the temperature is rising fairly quickly, the algorithm will respond more aggressively to prevent it from going over the limit for any appreciable amount of time. If the temperature is rising fairly slowly, however, the algorithm will respond less aggressively, allowing the temperature to inch closer to the limit.
Ok, so let's pull this all together. Knowing that the processor can run without issue at any temperature up to its Tcontrol temperature, the fan speed control algorithms will simply allow it to do so. This is why you see temperatures rising into the 80's as the Tcontrol thresholds are typically in the low- to mid-80's. Most people, in their ignorance, get worried when they see the temperatures getting higher and the fan not responding. There is nothing to be worried about, however. When the processor fan actually needs to respond, it will do so - and, because of the Output Weighting Matrix, it can also use the Chassis fans to assist in this temperature reduction when necessary.
That's the high-level introduction to the capability. If you would like, I can talk a bit about how you can use the configuration capabilities provided in BIOS Setup (Visual BIOS) to modify the default response. Unfortunately, it appears that this information was not properly documented in the BIOS Glossary or board TPS as it should have been (sigh!) and I will need to reconstitute it from memory (double sigh!). I warn you, this is a very involved process, with a *huge* number of control parameters, and it will likely take me some time to put this together...
Thanks for the thorough explanation on how the fan sub-system works Scott! I actually found what I was looking for, and indeed it was just a setting that I had missed, though it could have been a little more obvious (or as you mentioned, documented somewhere!).
The min/max duty cycle of the fans is under the sub-menu of the fan itself on the Cooling tab, but the target CPU temperature information is held under (duh) the sub-menu of the CPU Core Temp (or whatever target you want to focus on) just a bit below. Clicking on that temperature brings up Over-temp Threshold, Control Temperature (the target CPU temperature I was looking for), and All-On Temperature. After tweaking the Control Temperature, things react exactly how one might expect, though one of the intake fans isn't quite in line with the other (I blame the voltage control vs the PWM control, as well as the age of the fan). As you suspected, the Control Temperature value was in the mid 80's, which lines up with the fan curves that I wasn't seeing.
I know that manufacturers are perfectly comfortable with temperatures creeping that high, and though it's certainly hard to argue with the folks that make them, I do in my "ignorance" like to keep things a bit lower. Sacrificing a slight bit of fan noise (this is an S SKU after all, with large diameter fans) is a very small price to pay in order to keep things up to a temperature where I can comfortably touch the heatsink, in that infrequent time where it will have a long enough workload to bring it that high. Especially on an old(er) piece of equipment that would be difficult to replace, the peace of mind is worth those few DBs! I'm just thankful that the option is available, as opposed to more closed systems (an HP workstation, for example) where there's no option to modify that target value.