Proteinbenchmark | JM will implement limited rcloneing and not copy over the entire results directory JM – Done
JM will implement request for 20GB ephemeral storage. JM – Done
JM will adapt YAML template and individual umbrella window script to pass FF, Window, Replica, Target etc directly through environment variables rather than through the intermediary of CLI arguments JM – Done. CC – No issues here as far as I can tell.
CC update CC – Launched 20 jobs yesterday and ecerything went fine, and 40 jobs today. So seems to be working as planned. Haven’t tries convntional MD runs yet, but based on JM’s changes I should be able to do this once my current runs finish.
JW – Utilization looks good. LW’s jobs had just finished when I woke up and were idling a bunch of GPUs so I turned them off. But these all look great. JM, please offer to help LW with that (even knowing that evaluator is more complicated) JM – Will do
JW – Further things to do? JM – I’ve implemented everything I could think of. CC – This seems to be working well. One potential change on my end would be to break up the trajectories, but I can do that in the repo. Not a blocker. JW – Sounds good, feel free to request more GPUs JM – It looks like there are lots unused (the “gpumon” user is likely unused GPUs)
|