News for hackers. Stuff that matters.
Re:Building a supercomputer from 64 Raspberry Pis and Lego by bravura on Wednesday September 12, 2012 @05:50PM (#4513439)
I programmed MPI about 15 years ago, during a summer at the MIT AI Lab.

I implemented neural network feature creation for a backgammon agent ("Automated feature selection to maximize learning in artificial intelligence").

Nowadays, I mainly do parallel machine learning on machines with higher network latency. I haven't used MPI since.

Re:Building a supercomputer from 64 Raspberry Pis and Lego by apawloski on Wednesday September 12, 2012 @06:32PM (#4513619)
What do you use now instead of MPI?
Re:Building a supercomputer from 64 Raspberry Pis and Lego by pm90 on Thursday September 13, 2012 @01:51AM (#4514956)
He talked about higher latency so I'm guessing just sockets
Re:Building a supercomputer from 64 Raspberry Pis and Lego by bravura on Thursday September 13, 2012 @04:46AM (#4515299)
Incorrect.

With the sort of work I do nowadays (large-scale ML and NLP), I generally need very little synchronization, i.e. my tasks are usually embarrassingly parallel. I typically save final results in a centralized store (DB or NFS) and look at it there.

I also use Hadoop where appropriate.