Item | Notes |
|---|
Updates from MolSSI | BP – I reprovisioned the gateway server to become the production server last Friday. There were a few minutes of downtime but I think it has worked. BP – We’ve submitted a proposal for a new server. It’s an internal instrumentation proposal, so we need to justify how it would aid internal research and lead to publications. JW – We’d be happy to share authorship positions on mainline OpenFF force fields with MOlSSI/VT staff. And the OpenMM/Chodera lab ML datasets are probably in the same boat. DD – AFAIK, Chodera lab/OpenMM doesn’t have an alternative for the scale of QM compute they want for ML work. BP – This dataset “NABLA-DFT” is 7 TB - 1 million molecules, 5 million confs, WITH hamiltonians (which we generally can’t do). We can make the case that this would accelerate VT research/garner publications. Asking for $100k. Current proposal is for 100-200TB, with RAID so user would see 70TB available. And 60 cores. DD – Also, will want an export pathway for archival/to free up space. BP – At this amount of space, we could use the extra space for HDF5 storage, but eventually we’ll want to release these to Zenodo. BP – Also, I asked about using spinning disks, and was told that it doesn’t make a lot of sense, so instead we’re provisioning with SSDs. But we’ll be able to connect to external storage. JW will work with governing board to see how he can support getting the equipment supplement submitted by March.
|
Infrastructure advances | |
Throughput status | SPICE DES370K Single Points Dataset v1.0 OpenFF Protein Capped 3-mer Backbones v1.0 311691 == 311691, no change, 22/54 TDs Moved to scientific review
OpenFF Protein Capped 1-mer Sidechains v1.2 155215 == 155215, no change,, 45/46 TDs Moved to scientific review
OpenFF PEPCONF OptimizationDataset v1.0 2892 → 3074 Some segmentation faults JW – Is 200/week a good pace, or has this just hit the next roadblock? (Looks like about 200 completes, 200 errors, so we’ll leave this on)
|
User questions/issues | |
Science needs | |