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 
With Lambda and S3 we can get around that by dynamically allocating both compute and memory, and exploiting the insane S3 bandwidth (up to 500 GB/sec, not a typo) to be more efficient in terms of compute resources while being slightly slower to run end-to-end /3
Now your response might be like Dr Malcom's, and I'm sympathetic! But I think it's incredibly exciting to be pushing stateless-services (with backing stateful stores of incredible scale) in this direction, and I really do dream of a day of fully-elastic linear algebra. /4
Together, we can Make SVD Great Again! And if Moore's law really is dying, more efficient use of compute is going to become all the more vital in the future. /5
