Publications
Detailed Information
Dopia: Online Parallelism Management for Integrated CPU/GPU Architectures
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Cho, Younghyun | - |
dc.contributor.author | Park, Jiyeon | - |
dc.contributor.author | Negele, Florian | - |
dc.contributor.author | Jo, Changyeon | - |
dc.contributor.author | Gross, Thomas R. | - |
dc.contributor.author | Bernhard, Egger | - |
dc.date.accessioned | 2022-06-23T03:59:42Z | - |
dc.date.available | 2022-06-23T03:59:42Z | - |
dc.date.created | 2022-05-09 | - |
dc.date.issued | 2022-04 | - |
dc.identifier.citation | Proceedings of the ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPOPP, pp.32-45 | - |
dc.identifier.uri | https://hdl.handle.net/10371/183702 | - |
dc.description.abstract | © 2022 ACM.Recent desktop and mobile processors often integrate CPU and GPU onto the same die. The limited memory bandwidth of these integrated architectures can negatively affect the performance of data-parallel workloads when all computational resources are active. The combination of active CPU and GPU cores achieving the maximum performance depends on a workload's characteristics, making manual tuning a time-consuming task. Dopia is a fully automated framework that improves the performance of data-parallel workloads by adjusting the Degree Of Parallelism on Integrated Architectures. Dopia transparently analyzes and rewrites OpenCL kernels before executing them with the number of CPU and GPU cores expected to yield the best performance. Evaluated on AMD and Intel integrated processors, Dopia achieves 84% of the maximum performance attainable by an oracle. | - |
dc.language | 영어 | - |
dc.publisher | Association for Computing Machinery | - |
dc.title | Dopia: Online Parallelism Management for Integrated CPU/GPU Architectures | - |
dc.type | Article | - |
dc.identifier.doi | 10.1145/3503221.3508421 | - |
dc.citation.journaltitle | Proceedings of the ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPOPP | - |
dc.identifier.scopusid | 2-s2.0-85127579927 | - |
dc.citation.endpage | 45 | - |
dc.citation.startpage | 32 | - |
dc.description.isOpenAccess | N | - |
dc.contributor.affiliatedAuthor | Bernhard, Egger | - |
dc.type.docType | Conference Paper | - |
dc.description.journalClass | 1 | - |
- Appears in Collections:
- Files in This Item:
- There are no files associated with this item.
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.