Thanks Uplinger, this sounds like a wonderful idea. I am not having any issues getting tasks

Uplinger, how does resizing affect checkpointing? Got 2 running that had not done so for 4:50 hours, and this on a laptop. Less dockings, if that is what you're planning to do is not going to change that. It's a críme running these even size them down to just one job, task being a poor word choice in the log, in a task for the tablet. Unplug and poof 8-10 hours are gone. Have had multiple running 25-30 hours with 3 tasks in them according the result log, now one at 80% at 24:51 hours, and analysis shows getting 95% efficiency on average. Either the molecules to dock are made lighter for the androids to process or mid job interval checkpoint is introduced, not really appealing there large models being saved to internal memory on frequent basis. Something's gotta give, as a nightly on-charger crunch produces next to nothing now, a wholly unsatisfactory situation, not to speak of the pile of 195 fails seen on wingman, even my device has started to show them in series at times, all 4 fahv threads bombing.

Now running 7.3.17 and seen a signal 9, heartbeat and retry running to end and sitting in pending validation. No wingman yet that succeeded, so can't say if it's going the pending verification route and the eventual invalid.

So much for an off topic, on topic being, is it because the total work processed on the grid decreasing allowing you to increase share or is it that you've resolved the storage space and passthrough issue?

lavaflow,

Android does allow for mid job checkpoints. We will just be decreasing the average runtime. So if a work unit was estimated to take 5 hours, we will change that to 4 hours. This means putting a smaller number of jobs in a work unit.

As for bringing up the weight of MCM1, the researchers have taken steps to keep the size of the results lower. The decrease in the grid runtime overall does help, but runtime was not the driving factor for the file size issue.

Thanks,
-Uplinger

Btw, if you have mastered mid job checkpointing, fahv is meant, is this going to get ported to the windows and other platform versions? That could a peace many.

Thanks for the expansion. You'd be the winner getting mid-dock saves. Reading around, see you face resume issues, recreating the model and then compute on and then have this pass the validator and a quorum matching, if quorum is required. Right now, both tablet and laptop are off wcg as there's nothing to compute in fair time to even get one task a day to complete in a 'don't want to babysit' ambient. We should be free of concerns loosing umpteen hours of progress because we want to pick up and leave or power down for the devices not having to be on. Yes, all my devices still have a power switch, for the tongue in cheek.