Dr. Stephen Sanderson, University of Queensland

Dr. Stephen Sanderson is a postdoctoral research fellow in Prof. Debra Bernhardt’s group at the Australian Institute for Bioengineering and Nanotechnology (AIBN), The University of Queensland, where his studies focus on the theory behind computational methods for and statistical mechanics of non-equilibrium systems. Prior to that, he completed his Ph.D. at James Cook University under the supervision of Prof. Ron White and Dr. Bronson Philippa, where he investigated charge transport and exciton dynamics in organic semiconducting materials using kinetic Monte-Carlo and molecular dynamics techniques.

Title: Making the work-efficient parallel prefix sum do less work

Abstract:  The prefix sum (cumulative sum) algorithm has been accelerated for parallel processing through various algorithms, including the work-efficient algorithm in which the calculation is performed in “up sweep” and a “down sweep” stages. This talk begins by discussing details of implementing and optimizing this algorithm on GPU, as outlined in [1]. A common use of the prefix sum is to aid in selecting a random element from an array of probabilities. The cumulative sum of the array is generated, and then an element is chosen such that it is the last element with a value lower than or equal to a random number. In this talk, modifications to the work-efficient parallel prefix sum algorithm and the subsequent search algorithm are described which avoid unnecessary work in this specific “sum and search” use case.

[1] M. Harris, S. Sengupta, J. Owens; Chapter 39: Parallel Prefix Sum (Scan) with CUDA, GPU Gems 3 (2007), Available online: https://developer.nvidia.com/gpugems/gpugems3/part-vi-gpu-computing/chapter-39-parallel-prefix-sum-scan-cud

Dr. Jack Yang, University of New South Wales

Abstract: