Quote from: joseph32 on May 22, 2019, 07:29:30 PM

I see something strange with the newest 0.11.0:

This only happens to 1080 Ti, but on all of my rigs. They are slightly overclocked, but 100% stable since over 1 year. Every 5-15 hours (completely random) 1-2 cards are fully crashing and then they look like deactivated. The watchdog want to restart the miner in a loop, but cant because of the "deactivated" cards. I have to reboot the system and everything is fine again. (If they would crash for real, these cards would have the stock clocks after reboot, but they still have all the settings. So its definitely not a real crash of that cards for sure.)

I dunno when and why, its completely random. But this does not happen with 0.10.2. Switched few rigs over and they run fine since 2 days without any failure.
Any guess what this could be? Already testing the intensity, already switched off the watchdog, but right now waiting for the next crash. But maybe you already have an idea.

Same thing with my 1070s: one rig died this very same way on SMOS, t-rex-v0.11.0-cuda9.2 - had to re-flash SMOS to bring it working again

Also there is a bug with t-rex-v0.11.0-cuda9.2 (in different SMOS versions): when i change overclockig settings and restart miner program it starts hashing in half speed literally. Only rig reboot brings back hashing speed, miner program restart doesn't help.