Jeremy Kepner emailed yesterday to share some exciting news from MIT’s Lincoln Labs’ TX-2500, part of the LLGrid interactive/on-demand parallel computing system.
MIT Lincoln Laboratory (www.ll.mit.edu) has demonstrated a 0.5 Petabyte calculation on the HPC Challenge Stream benchmark (www.hpcchallenge.org) using its TX-2500 computer and pMatlabXVM software. This is the largest single problem ever run on a computer.
The TX-2500 has 432 dual core Xeon IB-connected nodes and 3.4TB of memory. I’ve emailed Dr. Kepner for clarification on how they know it’s the largest program ever run…
Anyway, the other interesting part of this is that the calculation used Matlab. LLGrid supports 200 ish users at Lincoln Lab, about 85% of whom run parallel Matlab codes.
I couldn’t find the entire release posted online in a quick google of the interwebs, so here it is if you’d like more detail.