Name
Abstract | ||
|---|---|---|
Given current hardware trends, ExaScale computing (1018 floating point operations per second) is projected to be available in less than a decade, achieved by using a huge number of processors, of order 109. Given the likely hardware heterogeneity in both platform and network, and the possibility of intermittent failures, dynamic scheduling will be needed to adapt to changing resources and loads. This will make it likely that repeated runs of a program will not execute operations like reductions in exactly the same order. This in turn will make reproducibility, i.e. getting bitwise identical results from run to run, difficult to achieve, because floating point operations like addition are not associative, so computing sums in different orders often leads to different results. Indeed, this is already a challenge on today's platforms. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1109/ARITH.2013.43 | IEEE Symposium on Computer Arithmetic |
Keywords | Field | DocType |
numerical reproducibility,dynamic scheduling,current hardware trend,exascale computing,likely hardware heterogeneity,different order,bitwise identical result,floating point operation,huge number,different result,repeated run,accuracy,hardware,floating point arithmetic,shape,parallel processing,computational modeling,addition | Exascale computing,Associative property,Bitwise operation,FLOPS,Computer science,Floating point,Parallel computing,Parallel processing,Theoretical computer science,Dynamic priority scheduling | Conference |
ISSN | Citations | PageRank |
1063-6889 | 8 | 0.58 |
References | Authors | |
3 | 2 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| James Demmel | 1 | 4817 | 551.47 |
| Hong Diep Nguyen | 2 | 138 | 8.93 |