I just built a computer with the Supermicro M12SWA-TF motherboard, for the WRX80 Threadripper Pro platform. It comes with a BMC and IPMI.
I discovered that when I put the computer to sleep and wake it up, all of the temperature and fan sensors return null values. In addition, the fans all spin at partial speed. I don't know the precise RPM, since the sensor values are null, but my peak load CPU temperature goes from 77C to 90C after waking from sleep.
Rebooting the computer reactivates the sensors, and returns the peak load CPU temperature to 77C.
If you put the computer to sleep, and thereby deactivate the sensors, and then hibernate the computer, the sensors reactivate to normal. So hibernation - by resetting the hardware but not Windows - successfully resets the sensors, even if the sensors were null at the time of hibernation. So to me, this indicates that the problem is with the motherboard, not Windows (since hibernation resets the hardware but not the software).
I reported this issue to Supermicro, and they said this is normal. The IPMI activates the sensors during bootup. Putting the computer to sleep deactivates the sensors, and they can't be reactivated without going through POST/BIOS.
Supermicro said that this is why they sell variants of a given motherboard with and without a BMC. The BMC is meant for a remote server that never sleeps, while the lack of a BMC is meant for an under-the-desk workstation that periodically sleeps.
This particular motherboard is only offered with a BMC. Likewise, Asus and Gigabyte's WRX80 motherboards all have a BMC.
So I've just accepted that I'll use hibernation instead of sleep. To be safe, I removed the sleep option from Windows, lest I accidentally sleep and fry my CPU.
But my question is, is this really normal? Do BMC/IPMI-equipped motherboards from companies besides Supermicro also disallow sleep / malfunction when sleep is used? Is this really normal?
I discovered that when I put the computer to sleep and wake it up, all of the temperature and fan sensors return null values. In addition, the fans all spin at partial speed. I don't know the precise RPM, since the sensor values are null, but my peak load CPU temperature goes from 77C to 90C after waking from sleep.
Rebooting the computer reactivates the sensors, and returns the peak load CPU temperature to 77C.
If you put the computer to sleep, and thereby deactivate the sensors, and then hibernate the computer, the sensors reactivate to normal. So hibernation - by resetting the hardware but not Windows - successfully resets the sensors, even if the sensors were null at the time of hibernation. So to me, this indicates that the problem is with the motherboard, not Windows (since hibernation resets the hardware but not the software).
I reported this issue to Supermicro, and they said this is normal. The IPMI activates the sensors during bootup. Putting the computer to sleep deactivates the sensors, and they can't be reactivated without going through POST/BIOS.
Supermicro said that this is why they sell variants of a given motherboard with and without a BMC. The BMC is meant for a remote server that never sleeps, while the lack of a BMC is meant for an under-the-desk workstation that periodically sleeps.
This particular motherboard is only offered with a BMC. Likewise, Asus and Gigabyte's WRX80 motherboards all have a BMC.
So I've just accepted that I'll use hibernation instead of sleep. To be safe, I removed the sleep option from Windows, lest I accidentally sleep and fry my CPU.
But my question is, is this really normal? Do BMC/IPMI-equipped motherboards from companies besides Supermicro also disallow sleep / malfunction when sleep is used? Is this really normal?