We perform Singular Value Decomposition (SVD) calculations on large datasets.
We modify the computation both by using fully precise and approximate methods, and by using both CPUs and GPUs.
In the end we compute an approximate SVD of 200GB of simulated data and using a mutli-GPU machine in 15-20 seconds.
Then we run this from a dataset stored in the cloud where we find that I/O is, predictably, a major bottleneck.
V. John. Scientific Computing and Applications, volume 7 of Advances in Computation: Theory and Practice, page 75+. Huntington, New York, Nova Science Publishers, Inc., (2001)
I. Babuska, A. Craig, J. Mandel, and J. Pitkäranta. BN-1105. Institute for Physical Science and Technology, Universityof Maryland, College Park, MD, (October 1989)
M. Fournié, N. Renon, Y. Renard, and D. Ruiz. Euro-Par 2010 - Parallel Processing, volume 6272 of Lecture Notes in Computer Science, Springer, Berlin, (2010)