Publications
Detailed Information
Dataflow Mirroring: Architectural Support for Highly Efficient Fine-Grained Spatial Multitasking on Systolic-Array NPUs
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Lee, Jounghoo | - |
dc.contributor.author | Choi, Jinwoo | - |
dc.contributor.author | Kim, Jaeyeon | - |
dc.contributor.author | Lee, Jinho | - |
dc.contributor.author | Kim, Youngsok | - |
dc.date.accessioned | 2024-05-02T05:42:40Z | - |
dc.date.available | 2024-05-02T05:42:40Z | - |
dc.date.created | 2024-04-23 | - |
dc.date.created | 2024-04-23 | - |
dc.date.issued | 2021 | - |
dc.identifier.citation | 2021 58TH ACM/IEEE DESIGN AUTOMATION CONFERENCE (DAC), pp.247-252 | - |
dc.identifier.issn | 0738-100X | - |
dc.identifier.uri | https://hdl.handle.net/10371/200475 | - |
dc.description.abstract | We present dataflow mirroring, architectural support for low-overhead fine-grained systolic array allocation which overcomes the limitations of prior coarse-grained spatial-multitasking Neural Processing Unit (NPU) architectures. The key idea of dataflow mirroring is to reverse the dataflows of co-located Neural Networks (NNs) in horizontal and/or vertical directions, allowing allocation boundaries to be set between any adjacent rows and columns of a systolic array and supporting up to four-way spatial multitasking. Our detailed experiments using MLPerf NNs and a dataflow-mirroring-augmented NPU prototype which extends Google's TPU with dataflow mirroring shows that dataflow mirroring can significantly improve the multitasking performance by up to 46.4%. | - |
dc.language | 영어 | - |
dc.publisher | IEEE | - |
dc.title | Dataflow Mirroring: Architectural Support for Highly Efficient Fine-Grained Spatial Multitasking on Systolic-Array NPUs | - |
dc.type | Article | - |
dc.identifier.doi | 10.1109/DAC18074.2021.9586312 | - |
dc.citation.journaltitle | 2021 58TH ACM/IEEE DESIGN AUTOMATION CONFERENCE (DAC) | - |
dc.identifier.wosid | 000766079700042 | - |
dc.identifier.scopusid | 2-s2.0-85109948613 | - |
dc.citation.endpage | 252 | - |
dc.citation.startpage | 247 | - |
dc.description.isOpenAccess | N | - |
dc.contributor.affiliatedAuthor | Lee, Jinho | - |
dc.type.docType | Proceedings Paper | - |
dc.description.journalClass | 1 | - |
- Appears in Collections:
- Files in This Item:
- There are no files associated with this item.
Related Researcher
- College of Engineering
- Department of Electrical and Computer Engineering
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.