2018
DOI: 10.1007/978-3-319-78024-5_16
|View full text |Cite
|
Sign up to set email alerts
|

TaskUniVerse: A Task-Based Unified Interface for Versatile Parallel Execution

Abstract: Task based parallel programming has shown competitive outcomes in many aspects of parallel programming such as efficiency, performance, productivity and scalability. Different approaches are used by different software development frameworks to provide these outcomes to the programmer, while making the underlying hardware architecture transparent to her. However, since programs are not portable between these frameworks, using one framework or the other is still a vital decision by the programmer whose concerns … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
3
0

Year Published

2019
2019
2019
2019

Publication Types

Select...
1
1
1

Relationship

2
1

Authors

Journals

citations
Cited by 3 publications
(4 citation statements)
references
References 21 publications
0
3
0
Order By: Relevance
“…This is an unnecessary obstacle to the ease of programming in the framework. Therefore, in an ongoing project, we are developing a unified interface, such that the algorithm and the task kernels only need to be defined once and then can be reused for all levels [39].…”
Section: Discussionmentioning
confidence: 99%
See 1 more Smart Citation
“…This is an unnecessary obstacle to the ease of programming in the framework. Therefore, in an ongoing project, we are developing a unified interface, such that the algorithm and the task kernels only need to be defined once and then can be reused for all levels [39].…”
Section: Discussionmentioning
confidence: 99%
“…Ideally, they should be the same in a general L-level framework. A unified interface that allows different run-times to cooperate transparently using a single application code is described, implemented and shown to perform well in [39].…”
Section: Taskifying Algorithmsmentioning
confidence: 99%
“…From the user perspective, it would be ideal if there were only one framework for task parallelism, or at least one common standard for task parallel programming implemented by different frameworks. Steps are being taken in this direction, see also the implementation in [62], but it will take some time until it is in place. Meanwhile, we provide an overview of some of the more relevant initiatives.…”
Section: A Task-parallel Programming Overviewmentioning
confidence: 99%
“…DuctTeip naturally supports multiple versions of the same data, which eliminates write-after-write dependencies. In a later work the authors of DuctTeip propose TaskUniVerse [190] as a general-purpose Application Programming Interface (API) that can serve as an abstraction layer for several tasking back-ends, such as cpuBLAS, SuperGlue, StarPU, DuctTeip or cuBLAS.…”
Section: Distributed Memory Tasksmentioning
confidence: 99%