I'd like to know it this behaviour is caused by the 1.39f5 SMC version or by the Base_XX blob in the NVRAM, what I call hardware descriptor. Apple continuously upgraded the Base_XX, from Base_17 in February 2009 to Base_21 in the end of 2011.
If you wanna troubleshot this at a lowest level possible, I can upgrade your Base_XX to the same one as the mid-2012 MP51 (Base_21). Just do a dump of your BootROM, compress it and send me by PM.
@h9826790 did you had this bug too? Resolved with Base_21?
Sure I can monitor it. I had this bug for many years.
I have to go to work soon. So, can't test it properly now. But will try to disable the fan control later and see if the updated Base_XX fix it.
However, since this bug is an intermittent bug. And can be temporarily fixed by SMC reset, or even just keep rebooting the cMP. The “firmware update” process itself may be good enough to temporarily fix it, and hide the fact.
Therefore, it will need may be weeks to confirm if that’s a real possible fix. Of course, if nothing fixed, we can know that straight away.
Anyway, for leon771, a quick work around is just stress the GPU a little bit (e.g. run Luxmark for few seconds), then the SMC will back to normal, the PSU / PCIe fan can go back to their real idle speed. You and make Luxmark auto run on every login, and then manually close it after a few seconds rendering.
Or, you can use MacsFanControl to setup a fan profile for PCIe and PSU fan. It can also fix the issue. I use my own fan control software to manage the PCIe and PSU fan automatically via modifying their target fan speed (base on PCIe ambient temperature and PSU component temperature). MacsFanControl can do the same thing.
The native PSU and PCIe fan profile base on both power draw and temperature. However, spin up very aggressively when PCIe power draw increase, but too gentle when temperature increases. For me, that’s a very strange fan profile. So, regardless if there is a bug, I will make my own fan profile for them anyway.
IMO, the fan should react to the actual temperature more. The native fan setting now may lead to something like
1) fan spin up a lot when just start to tax the GPU (when it is still quite cool), but spin down too soon when the GPU still hot (due to power draw decrease when the stress level goes down).
2) fan spin unnecessary fast in cold winter due to high power draw when under stress, even actual temperature isn't that high.
3) PSU high temperature due to high usage from components other than GPU (or PCIe cards)
Anyway, I will check if my PCIe and PSU fan back to normal later, and report back.