Hoping to "look upon the snow with new eyes". You're not full-stack unless you're worried about convergence and Johnson noise. Very happy postdoc.
May 27, 2020 • 12 tweets • 2 min read
How would I improve American scientific output for $100B? A thread. 1/n
First, we already have a lot of processes that seem to work incredibly well, and processes really matter! Institutional knowledge matters! So whatever you do, _don't break existing things_. 2/n
Oct 24, 2018 • 6 tweets • 3 min read
Announcing our work on building a Serverless Supercomputer: NumPyWren. Can you get Lambda to do large-scale linear algebra with S3 as your memory? arxiv.org/pdf/1810.09679… The answer is an emphatic yes! For some algorithms, it's even more resource-efficient than MPI @ucbrise /1
How can this be? Many HPC algorithms exhibit steadily-declining working-set size and dampened-sinusoid-like patterns of parallelism. A model which colocates compute and storage ends preallocates them can end up with a lot of idle resources sitting around. /2
