Abstract | ||
---|---|---|
Floating-point additions in concurrent execution environment are known to be hazardous, as the result depends on the order in which operations are performed. This problem is encountered in data parallel execution environments such as GPUs, where reproducibility involving floating-point atomic addition is challenging. This problem is due to the rounding error or cancellation that appears for each operation, combined with the lack of control over execution order. In this article we propose two solutions to address this problem: work reassignment and fixed-point accumulation. Work reassignment consists in enforcing an execution order that leads to weak reproducibility. Fixed-point accumulation consists in avoiding rounding errors altogether thanks to a long accumulator and enables strong reproducibility. |
Year | DOI | Venue |
---|---|---|
2015 | 10.15439/2015F86 | 2015 Federated Conference on Computer Science and Information Systems (FedCSIS) |
Keywords | Field | DocType |
reproducible floating-point atomic addition,data-parallel environment,concurrent execution environment,data parallel execution environments,GPU,rounding error cancellation,work reassignment,fixed-point accumulation,execution order,weak-reproducibility | Kernel (linear algebra),Synchronization,Computer science,Floating point,Instruction set,Round-off error,Parallel computing,Rounding,Accumulator (structured product) | Conference |
Volume | ISSN | Citations |
5 | 2300-5963 | 1 |
PageRank | References | Authors |
0.40 | 8 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
David Defour | 1 | 131 | 18.28 |
Sylvain Collange | 2 | 14 | 2.58 |