Name
Abstract | ||
|---|---|---|
Previous parallel sorting algorithms do not scale to the largest available machines, since they either have prohibitive communication volume or prohibitive critical path length. We describe algorithms that are a viable compromise and overcome this gap both in theory and practice. The algorithms are multi-level generalizations of the known algorithms sample sort and multiway mergesort. In particular, our sample sort variant turns out to be very scalable both in theory and practice where it scales up to 215 MPI processes with outstanding performance in particular for medium sized inputs. Some tools we develop may be of independent interest -- a simple, practical, and flexible sorting algorithm for very small inputs, a near linear time ptimal algorithm for solving a constrained bin packing problem, and an algorithm for data delivery, that guarantees a small number of message startups on each processor. |
Year | DOI | Venue |
|---|---|---|
2015 | 10.1145/2755573.2755595 | ACM Symposium on Parallelism in Algorithms and Architectures |
Keywords | Field | DocType |
parallel sorting,multiway mergesort,sample sort | Merge algorithm,Counting sort,Sorting network,Hybrid algorithm,Computer science,Parallel computing,External sorting,Selection sort,Bucket sort,Sorting algorithm,Distributed computing | Conference |
Citations | PageRank | References |
9 | 0.76 | 27 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Michael Axtmann | 1 | 22 | 3.72 |
| Timo Bingmann | 2 | 24 | 5.83 |
| peter sanders | 3 | 361 | 29.35 |
| Christian Schulz | 4 | 240 | 24.10 |