Using Butterfly-Patterned Partial Sums to Optimize GPU Memory Accesses for Drawing from Discrete Distributions
Oracle Labs
arXiv:1505.03851 [cs.DC], (14 May 2015)
@article{.2015using,
title={Using Butterfly-Patterned Partial Sums to Optimize GPU Memory Accesses for Drawing from Discrete Distributions},
author={., Guy L. Steele Jr},
year={2015},
month={may},
archivePrefix={"arXiv"},
primaryClass={cs.DC}
}
We describe a technique for drawing values from discrete distributions, such as sampling from the random variables of a mixture model, that avoids computing a complete table of partial sums of the relative probabilities. A table of alternate ("butterfly-patterned") form is faster to compute, making better use of coalesced memory accesses. From this table, complete partial sums are computed on the fly during a binary search. Measurements using an NVIDIA Titan Black GPU show that for a sufficiently large number of clusters or topics (K > 200), this technique alone more than doubles the speed of a latent Dirichlet allocation (LDA) application already highly tuned for GPU execution.
May 16, 2015 by hgpu
