Teamredminer reboots when GPU Mem drops to 100 MHz

[afargo]

I have two 5700xt rigs and one of them experiences multiple teamredminer reboots every day. The reboots do not seem to be associated with any particular GPU (I’ve tried switching them out and it didn’t help). Looking through the logs, it appears that the Memory clock for one of the GPUs drops to 100 for some reason (I’ve got them set at 905-950 MHz) and the power consumption drops to between 34 and 53 Watts. I’ve attached a few screenshots.

Both rigs are running on 0.6-203 because higher versions are unstable. Any idea what could be causing the memory clock and power to crash on seemingly random GPUs?

[keaton_hiveon]

The cards are crashing either from the memory being set too high (likely) or the voltages being set too low. Adjust oc after each crash and reboot. Repeat until stable

[afargo]

My other 5700 rig has the same voltage settings, and almost identical memory clocks, and it never reboots. And the reboots randomly occur with GPUs that are set as low as 905, so I doubt that the memory clock is too high.

In my experience, if the memory clock is too high for a particular GPU, that GPU causes the system to reboot repeatedly. Here, none of the cards cause the system to reboot repeatedly. If one of them did, then I would adjust the memory clock down.

Also, neither of the GPUs that I moved from this rig to the other rig have caused reboots. And both of the GPUs that I swapped to this rig have caused reboots at the same OC settings as they were before, when they never caused reboots in the rig I moved them from.

[keaton_hiveon]

not all cards will clock the same, or run at the same settings forever.

reduce the mem and reboot. repeat until stable.