Docking one coreHPC

Docking on coreHPC/FAC

• to run docking at scale, you are doing as follows
• identify the subset to dock
• foreach tranche of 1 TB :
  • copy tranche to FAC
  • dock this tranche on coreHPC
  • repeat until 3000 tranches, all 3PB have been docked
• data transfer is expected to be 50 Gbps (pipe is 100 Gbps total)
• 1 TB at 50 Gbps takes 3 minutes.
• 3 PB takes 6.25 days.
• thus the process is probably CPU limited and not bandwidth limited.
• obviously work out a staging protocol while one tranche is transporting, one is docking, and one is being deleted.