DokuWiki

• Brief history of HPC
  • 2006 swallowtail (Dell PE1955, Infinband, imw, emw)
  • 2010 greentail (HP gen6 blades, hp12)
  • 2013 sharptail (storage, K20s, Infiniband, mw256/mwgpu))
  • 2014 mw256fd (Dell replacement)
  • 2015 tinymem (serial job expansion)
  • 2017 mw128 (first faculty startup funds)
  • 2018 Todays meeting

* Since 2006

• Grown from 256 to roughly 1,100 physical CPU cores
• Processed 3,165,752 jobs (18jun18)
• Compute capacity at 47-48 teraflops (DPFP)
• About 500 accounts created (incl 22 collaborators, 100 class accounts)

• Funding / charge scheme: is it working for you?
  • Last 2 years $15K target realized

• Status of our cluster development fund
  • $140K come July 1st, 2018
  • Time for some new hardware? Retirement of hp12 nodes?

• 2017 Benchmarks of some new hardware
  • Donation led to purchase of four GTX1080i commercial grade GPUs
  • Amber 16. Nucleosome bench runs 4.5x faster than on a K20
  • Gromacs 5.1.4. Colin's multidir bench runs about 2x faster than on a K20
  • Lammps 11Aug17. Colloid example runs about 11x faster than K20
  • FSL 5.0.10. BFT bedpostx tests run 16x faster on CPU, a whopping 118x faster on GPU.

• 2017 IBM purchased Platform Inc (developers of LSF, Openlava is LSF4.2 open source branch)
  • Promptly accused Openlava of copyright infringement in v3.x
  • Fall back option to v2.2 (definitely free of infringement, minor disruption)
  • Fall forward option, adopt SLURM (major disruption)

Back