Running a i4770 on a DH87RL board, with latest bios (320)
Using an PWM cooler from Noctua NH-12S (which received positive reviews from a number of magazines).
My problem: the mainboard doesn't seem to control the fan(s) properly.
When I use factory settings (setting "Quite(0) in cooling assistant) the CPU fan starts with about 300rpm and stays there. Case fase are running with 600-700 rpm.
When I put the CPU under stress, its temperature goes up to 80 C without the fans ever speeding up. Then the thermal throttle sets in.
With any different setting in cooling assistant (e.g. "Cooling Assistant(1)"), the fans start with full speed (1500rpm for CPU, >1000rpm for front and rear), even when the machine is idle (or is in the BIOS screen).
Now the CPU stays below 60C even under full load, but it's somewhat loud and I really would like to have proper thermal management.
RPM and temperature sensors seem to work correctly.
I noticed that the BIOS cooling assistant graphics seems inconsistent at times. If you adjust the temperature curves and switch to another tab and then back, the curves are not remembered correctly.
Is this a known BIOS issue?
the fan I'm talking about is the CPU fan and connected correctly. The observed behavior seems to cover all three fans, though (front, rear and CPU): they either stay at minimum and never speed up or they stay at maximum and never slow down.
And I'm saving after changing settings; the fan behavior did change after changing bios settings, just not in a way I'd expect.
I recommend you testing other another fan and also checking the heat sink. Most of the times when the system heats more than usual it is related to the Heat sink setup or the Thermal Material you put on.
Note: The information contained herein apply equally to all of the 8 Series Intel(R) Desktop Boards...
The Cooling Assistant offers you nine different automatic settings that you can choose from. You can choose to use one of these settings or you can go into the Advanced Cooling scene and manually set all of the parameters yourself.
The nine settings can be broken up into three sets of three. The three at the "Quiet" end of the spectrum (I will call them "Group A") utilize a PID algorithm, with Slow, Normal and Aggressive levels of responsiveness applied to it. The middle three (I will call them "Group B") utilize a piece-wise linear algorithm (again with Slow, Normal and Aggressive levels of responsiveness applied). Finally, the three at the "Cool" end of the spectrum (I will call them "Group C") also utilize the piece-wise linear algorithm, but in this case we (also) lower the control/limit temperature by an additional 10 degrees (below the processor's Tcontrol temperature).
In the case of the PID algorithm, it is tasked with keeping the temperature at or below the control temperature. By their very nature (you can look up PID algorithms online for more information), if the temperature rises very slowly, the algorithm will allow the temperature to rise all the way up to the control temperature without increasing the fan speed (this is what LimitingFactor is seeing). This is perfectly alright since the processor can run at its Tcontrol temperature constantly without any negative affects on its lifetime (we guarantee this). LimitingFactor is *NOT* seeing throttling as a result of this, however. The Tcontrol temperature is typically in the vicinity of 80c whereas the Tjmax temperature -- which is where thermal throttling is applied -- is typically in the vicinity of 100c. If the temperature goes above the Tcontrol temperature, the PID algorithm will aggressively apply fan speed to pull it back down (the higher the temperature, the more aggressive it will be). In addition, we set an All-On temperature threshold (typically 8 degrees above Tcontrol). If this threshold is exceeded, the fans are taken to 100%. Bottom line, it will not ever get to the vicinity of the Tjmax temperature -- nor result in any throttling occurring -- unless the physical cooling solution (i.e. heatsink and fan) is inadequate or improperly installed.
Many people find the PID algorithm a bit disconcerting -- or think it is not working properly -- because they don't understand it. They think that the higher temperatures are a problem, but they are not. Still, if you truly want to see a lower temperature maintained, you can choose from the Group B or Group C settings or you can switch to using the discrete thermal response and fan speed control parameters to choose an alternative. The PID algorithm offers a lot of other acoustic and psychoacoustic benefits. If you want to retain these benefits but want the temperature that is maintained to be lower, then use the discrete parameters to select the PID algorithm but lower the control temperature for the processor to something below its Tcontrol setting...
P.S. For the Group B and C automatic settings, there are other threads where I have detailed how the piece-wise linear algorithm works, so I didn't waste time detailing it here again. As I said earlier, if you don't like how any of the settings work, you can always choose to set the parameters discretely. In LimitingFactor's case, this is probably a good idea anyway, not because it isn't working properly but more because it sounds like the default minimum duty cycle setting for the fan need to be adjusted (increased) for this particular cooler anyway...
Try saving to a FAT32-formatted drive (flash stick for example). The file is being saved in BMP format, so you might want to include this extension in the file name (though this isn't absolutely necessary to save successfully).
As was related in another thread, an issue with the enable of the Piece-Wise Linear response algorithm - resulting in the fans all running at full speed - was reported against the DH87RL board. I have reproduced this issue on a DZ87KLT-75K board as well - which means that this issue will likely occur on any of the 8 Series boards.
I have since confirmed that there is indeed a bug in the fan speed control configuration code. The Intel(R) Quiet System Technology (QST) algorithms in the eSIO are being improperly programmed if either the Piece-Wise Linear or Simple Linear algorithms are selected. I am currently working on a fix. Updated releases of the BIOS for each board, as well as an updated release of Intel(R) Desktop Utilities (since it shares the same code), will be necessary to address the issue completely.
It will take a few weeks for the fix and the subsequent BIOS and IDU updates to be fully verified and released. In the meantime, if you are using the advanced cooling configuration parameters, stick with the Proportional-Integral-Derivative (PID) or Temperature Range Management algorithms. For those using only the Visual BIOS Cooling Assistant slider, this means using only settings 0, 1 or 2 for now...
Sorry for the inconvenience,
That is not unexpected. As I said, all of the 8 Series boards (DZ87KLT-75K, DH87RL, DH87MC, DQ87PG and DB85FL) are affected by this bug. The BIOSs providing the fix are in flight; watch for postings in the Download Center...
I have the same board as the topic starter and the same problem. Two weeks ago you have written, that the update is in flight. Can you be a bit more precise regarding to the release date, please?
And this:The Cooling Assistant offers you nine different automatic settings that you can choose from.
The Cooling Assistant can have ten positions inclusively the zero-position.
By the way I searched for information about the Temperature Range Management algorithms, but did not find anything. Would explain, how these algorithms work, please?
I think, that the position of the cooling assistant should change to "custom", whenever the user manually changes settings.
Sorry, I cannot commit a specific release date. All I can say is that the team is working diligently to deliver the BIOS updates as soon as possible (but there are a lot of updates in the works)...
Yes, the slider has ten positions, but only nine of them are used to select amongst the (nine) possible automatic modes.
The Temperature Range Management algorithm is fairly simple; it attempts to keep the temperature within a small range centered around the temperature specified by the Control Temperature parameter. While the temperature is within this range, the fan requirements are held where they are. When the temperature leaves this range, however, the duty cycle will be incremented or decremented as necessary to pull the temperature back into this range. How aggressively it does so is dependent upon the setting of the Responsiveness parameter. From a stability standpoint, this is the least stable of the algorithms. It has a tendency to cause noticeable fan oscillation, especially if you set the Responsiveness to High (aggressive). On the other hand, if your goal is to keep the temperature at some specific level - and you don't care about the acoustic or psycho-acoustic side-effects (perhaps you have really quiet fans and acoustics are thus not a concern) - then this is a reasonable choice. If you are going to try this algorithm, my recommendation is that you start with the Responsiveness parameter set to "Low" and increase it only if it really needs to respond more quickly (i.e. takes too long to drag the temperature back into the control range. Also, in the case of the Processor temperature, you will want to lower the Control Temperature away from the (default) Tcontrol level...
Yes, the Cooling Assistant's slider *should* move to the "Custom" position whenever the user manually makes changes to the temperature response and/or fan speed control settings. This is a bug in Visual BIOS. I have reported it to the team responsible...