fine grained bandwidth adaptivity in networks on chip
play

Fine-Grained Bandwidth Adaptivity in Networks-on-Chip Using - PowerPoint PPT Presentation

Mar 04, 2024 •20 likes •990 views

Fine-Grained Bandwidth Adaptivity in Networks-on-Chip Using Bidirectional Channels Robert Hesse , Jeff Nicholls, Natalie Enright Jerger University of Toronto May 10, 2012 Friday, 11 May, 12 Motivation May 10, 2012 2 University of Toronto

  1. Fine-Grained Bandwidth Adaptivity in Networks-on-Chip Using Bidirectional Channels Robert Hesse , Jeff Nicholls, Natalie Enright Jerger University of Toronto May 10, 2012 Friday, 11 May, 12

  2. Motivation May 10, 2012 2 University of Toronto Friday, 11 May, 12

  3. Motivation • NoCs are crucial for scaling CMPs May 10, 2012 2 University of Toronto Friday, 11 May, 12

  4. Motivation • NoCs are crucial for scaling CMPs • Problem: May 10, 2012 2 University of Toronto Friday, 11 May, 12

  5. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static Time May 10, 2012 2 University of Toronto Friday, 11 May, 12

  6. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic May 10, 2012 2 University of Toronto Friday, 11 May, 12

  7. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: May 10, 2012 2 University of Toronto Friday, 11 May, 12

  8. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: – Over-provisioned link BW May 10, 2012 2 University of Toronto Friday, 11 May, 12

  9. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: Average channel utilization: < 5% – Over-provisioned link BW May 10, 2012 2 University of Toronto Friday, 11 May, 12

  10. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: Average channel utilization: < 5% – Over-provisioned link BW • Our solution: May 10, 2012 2 University of Toronto Friday, 11 May, 12

  11. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: Average channel utilization: < 5% – Over-provisioned link BW • Our solution: – Adapt link BW to demands May 10, 2012 2 University of Toronto Friday, 11 May, 12

  12. Motivation • NoCs are crucial for scaling CMPs BW • Problem: – NoC bandwidth resources are static – Bandwidth requirements are highly Time dynamic • Current solution: Average channel utilization: < 5% – Over-provisioned link BW • Our solution: Save up to 75% of BW resources – Adapt link BW to demands May 10, 2012 2 University of Toronto Friday, 11 May, 12

  13. Motivation - Static NoC May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  14. Motivation - Static NoC IP IP IP IP 0 1 2 3 IP IP IP IP 4 5 6 7 IP IP IP IP 8 9 10 11 IP IP IP IP 12 13 14 15 • Static Topology May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  15. Motivation - Static NoC IP IP IP IP 0 1 2 3 IP IP IP IP 4 5 6 7 IP IP IP IP 8 9 10 11 IP IP IP IP 12 13 14 15 • Static Topology • Static Bandwidth May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  16. Motivation - Static NoC Uniform Random IP IP IP IP 0 1 2 3 IP IP IP IP 4 5 6 7 IP IP IP IP 8 9 10 11 IP IP IP IP 12 13 14 15 • Static Topology • Static Bandwidth • Static workloads for evaluation May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  17. Motivation - Static NoC Uniform Random 15 14 IP IP IP IP IP IP IP IP 13 12 0 0 1 1 2 2 3 3 11 10 IP IP IP IP IP IP IP IP 9 4 4 5 5 6 6 7 7 8 7 IP IP IP IP IP IP IP IP 6 8 8 9 9 10 10 11 11 5 4 3 IP IP IP IP IP IP IP IP 2 1 12 12 13 13 14 14 15 15 0 • Static Topology • Static Bandwidth • Static workloads for evaluation May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  18. Motivation - Static NoC Transpose 15 14 IP IP IP IP IP IP IP IP 13 12 0 0 1 1 2 2 3 3 11 10 IP IP IP IP IP IP IP IP 9 4 4 5 5 6 6 7 7 8 7 IP IP IP IP IP IP IP IP 6 8 8 9 9 10 10 11 11 5 4 IP IP IP IP IP IP IP IP 3 2 12 12 13 13 14 14 15 15 1 0 • Static Topology • Static Bandwidth • Static workloads for evaluation May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  19. Motivation - Static NoC Transpose 15 14 IP IP IP IP IP IP IP IP 13 12 0 0 1 1 2 2 3 3 11 10 IP IP IP IP IP IP IP IP 9 4 4 5 5 6 6 7 7 8 7 IP IP IP IP IP IP IP IP 6 8 8 9 9 10 10 11 11 5 4 IP IP IP IP IP IP IP IP 3 2 12 12 13 13 14 14 15 15 1 0 • Specified at design time • Static Topology for worst case scenario • Static Bandwidth • Static workloads for evaluation May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  20. Motivation - Static NoC Transpose 15 14 IP IP IP IP IP IP IP IP 13 12 0 0 1 1 2 2 3 3 11 10 IP IP IP IP IP IP IP IP 9 4 4 5 5 6 6 7 7 8 7 IP IP IP IP IP IP IP IP 6 8 8 9 9 10 10 11 11 5 4 IP IP IP IP IP IP IP IP 3 2 12 12 13 13 14 14 15 15 1 0 • Specified at design time • Static Topology for worst case scenario • Static Bandwidth • Static NoCs can handle • Static workloads for temporally- and spatially- evaluation stable traffic well May 10, 2012 University of Toronto 3 Friday, 11 May, 12

  21. Motivation - Real NoC Tra ffj c Blackscholes 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  22. Motivation - Real NoC Tra ffj c Blackscholes 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 • Highly dynamic workloads May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  23. Motivation - Real NoC Tra ffj c Blackscholes 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 • Highly dynamic workloads • Large temporal and spatial BW variance May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  24. Motivation - Real NoC Tra ffj c Streamcluster 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 • Highly dynamic workloads • Large temporal and spatial BW variance May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  25. Motivation - Real NoC Tra ffj c Streamcluster 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 • Highly dynamic • Significant area and power overhead with traditional workloads NoC implementation • Large temporal and spatial BW variance May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  26. Motivation - Real NoC Tra ffj c Streamcluster 15 IP IP IP IP 14 13 0 1 2 3 12 11 IP IP IP IP 10 9 4 5 6 7 8 7 IP IP IP IP 6 8 9 10 11 5 4 IP IP IP IP 3 2 12 13 14 15 1 0 • Highly dynamic • Significant area and power overhead with traditional workloads NoC implementation • Large temporal and • Channels are underutilized spatial BW variance most of the time May 10, 2012 University of Toronto 4 Friday, 11 May, 12

  27. Channel Utilization May 10, 2012 5 University of Toronto Friday, 11 May, 12

  28. Channel Utilization 45" 40" Max."U?liza?on" Channel'U)liza)on'(%)' 35" Avg."U?liza?on" 30" 25" 20" 15" 10" 5" 0" " " " " " " " " " " " s k l m t e s s . e r a g e e e n p c t c e v i l a r a t o i a o s n r s V A m r e r e ? u h t n t c F p y i y l c a a n c d a s a C F m a k o R w d c B a S a i u e l B l r F t S May 10, 2012 5 University of Toronto Friday, 11 May, 12

  29. Channel Utilization 45" 40" Max."U?liza?on" Channel'U)liza)on'(%)' 35" Avg."U?liza?on" 30" 25" 20" 15" 3.42% 10" 5" 0" " " " " " " " " " " " s k l m t e s s . e r a g e e e n p c t c e v i l a r a t o i a o s n r s V A m r e r e ? u h t n t c F p y i y l c a a n c d a s a C F m a k o R w d c B a S a i u e l B l r F t S May 10, 2012 5 University of Toronto Friday, 11 May, 12

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Fine-Grained Geographic Communication (Geocast) Nexus Workshop Frank Drr 23.07.2003 1

Fine-Grained Geographic Communication (Geocast) Nexus Workshop Frank Drr 23.07.2003 1

Fine-Grained Geographic Communication (Geocast) Nexus Workshop Frank Drr 23.07.2003 1 Overview Motivation Requirements for Fine-Grained Geocast Location Model for Fine-Grained Geographic Addressing Summary Related Work

752 views • 18 slides
Mechanized Verification of Fine-grained Concurrent Programs Ilya Sergey Aleks Nanevski

Mechanized Verification of Fine-grained Concurrent Programs Ilya Sergey Aleks Nanevski

Mechanized Verification of Fine-grained Concurrent Programs Ilya Sergey Aleks Nanevski Anindya Banerjee PLDI 2015 Terminology Coarse-grained Concurrency synchronisation between threads via locks ; Fine-grained

973 views • 72 slides
Parts of Speech More Fine-Grained Classes More

Parts of Speech More Fine-Grained Classes More

Parts of Speech More Fine-Grained Classes More Fine-Grained Classes Actually , I ran home extremely quickly yesterday The closed classes Example of POS tagging The Penn Treebank Part-of-Speech

747 views • 54 slides
Fine Grained Access Control Fine-Grained Access Control Fine Grained Access Control

Fine Grained Access Control Fine-Grained Access Control Fine Grained Access Control

Fine Grained Access Control Fine-Grained Access Control Fine Grained Access Control Fine-grained access control examples: p g Students can see their own grades Students can see grades of all students in courses they registered for

548 views • 34 slides
Fine-grained Visual Analysis: From Classification to Retrieval Yi-Zhe Song SketchX Lab, CVSSP,

Fine-grained Visual Analysis: From Classification to Retrieval Yi-Zhe Song SketchX Lab, CVSSP,

Fine-grained Visual Analysis: From Classification to Retrieval Yi-Zhe Song SketchX Lab, CVSSP, University of Surrey, UK http://sketchx.ai Why fine-grained? Dog Dog Dog I am not just a dog Why fine-grained? Husky

615 views • 24 slides
Exploring Chip to Chip Photonic Networks Philip Watts Computer Laboratory University of Cambridge

Exploring Chip to Chip Photonic Networks Philip Watts Computer Laboratory University of Cambridge

Exploring Chip to Chip Photonic Networks Philip Watts Computer Laboratory University of Cambridge Acknowledgement: Royal Commission for the Exhibition of 1851 Exploring Chip to Chip Photonics p g p p Bandwidth improvements in

90 views • 8 slides
Fine-grained Image Recognition Lei Wang VILA group School of Computing and Information

Fine-grained Image Recognition Lei Wang VILA group School of Computing and Information

Learning Kernel-matrix-based Representation for Fine-grained Image Recognition Lei Wang VILA group School of Computing and Information Technology University of Wollongong, Australia 11-Dec-2019 Introduction Fine-grained image recognition

1.56k views • 58 slides
TRILL Fine Grained Labeling Donald Eastlake 3 rd Huawei

TRILL Fine Grained Labeling Donald Eastlake 3 rd Huawei

TRILL Fine Grained Labeling Donald Eastlake 3 rd Huawei Technologies d3e3e3@gmail.com November 2011 TRILL WG 1 Fine Grained Labeling A way to

469 views • 8 slides
Fine-grained Transaction Scheduling in Replicated Databases via Symbolic Execution Pedro

Fine-grained Transaction Scheduling in Replicated Databases via Symbolic Execution Pedro

Fine-grained Transaction Scheduling in Replicated Databases via Symbolic Execution Pedro Raminhas pedro.raminhas@tecnico.ulisboa.pt Stage: 2 nd Year PhD Student Research Area: Dependable and fault-tolerant systems and networks Advisors: Miguel

662 views • 13 slides
Addressing Inter-Class Similarity in Fine-Grained Visual Classification Abhimanyu Dubey

Addressing Inter-Class Similarity in Fine-Grained Visual Classification Abhimanyu Dubey

Addressing Inter-Class Similarity in Fine-Grained Visual Classification Abhimanyu Dubey Collaborators: Nikhil Naik, Ryan Farrell, Otkrist Gupta, Pei Guo, Ramesh Raskar Fine-Grained Visual Classification - Image classification with target

760 views • 35 slides
On the Correctness Criteria of Fine-Grained Access Control in Relational Databases Qihua Wang,

On the Correctness Criteria of Fine-Grained Access Control in Relational Databases Qihua Wang,

On the Correctness Criteria of Fine-Grained Access Control in Relational Databases Qihua Wang, Ting Yu, Ninghui Li Jorge Lobo, Elisa Bertino Keith Irwin, Ji-Won Byun Outline Introduction Correctness Criteria A Fine-Grained Access

1.21k views • 51 slides
Fine-Grained Power Modeling for Smartphones Using System Call Tracing Based on paper and

Fine-Grained Power Modeling for Smartphones Using System Call Tracing Based on paper and

Fine-Grained Power Modeling for Smartphones Using System Call Tracing Based on paper and presentation by: Abhinav Pathak, Y. Charlie Hu, Ming Zhang Paramvir Bahl, Yi-Min Wang Damian Rodziewicz Fine-Grained Power Modeling for Smartphones

701 views • 33 slides
Fine-Grained Access Control Fine Grained Access Control Fine-grained access control examples:

Fine-Grained Access Control Fine Grained Access Control Fine-grained access control examples:

Fine-Grained Access Control Fine Grained Access Control Fine-grained access control examples: Students can see their own grades Students can see their own grades Students can see grades of all students in courses they registered for

338 views • 17 slides
Fine Grained Coordinated Parallelism in a Real World Application Mohammad Rezaei, PhD June 2012

Fine Grained Coordinated Parallelism in a Real World Application Mohammad Rezaei, PhD June 2012

Fine Grained Coordinated Parallelism in a Real World Application Mohammad Rezaei, PhD June 2012 1 Outline Types of parallelism Algorithm description Why fine grained parallelism? Concurrency is hard no, its different

419 views • 19 slides
Mesos A Platform for Fine-Grained Resource Sharing in the

Mesos A Platform for Fine-Grained Resource Sharing in the

Mesos A Platform for Fine-Grained Resource Sharing in the Data Center Benjamin Hindman, Andy Konwinski, Matei Zaharia , Ali Ghodsi, Anthony Joseph,

438 views • 32 slides
Enhancing Fine- Grained Parallelism Loop vectorization, Loop distribution, Scalar expansion

Enhancing Fine- Grained Parallelism Loop vectorization, Loop distribution, Scalar expansion

Enhancing Fine- Grained Parallelism Loop vectorization, Loop distribution, Scalar expansion Scalar and array renaming 1 Fine-Grained Parallelism Theorem 2.8. A sequential loop can be converted to a parallel loop if the loop carries no

606 views • 41 slides
The Zoltan Toolkit Partitioning, Ordering, and Coloring Erik Boman, Cedric Chevalier, Karen

The Zoltan Toolkit Partitioning, Ordering, and Coloring Erik Boman, Cedric Chevalier, Karen

The Zoltan Toolkit Partitioning, Ordering, and Coloring Erik Boman, Cedric Chevalier, Karen Devine Sandia National Laboratories, NM mit atalyrek Ohio State University Dagstuhl Seminar, Feb 2009 Sandia is a multiprogram laboratory

1.41k views • 100 slides
Capacity of Wireless Networks Anil Kumar S. Vullikanti Network Dynamics and Simulation Science

Capacity of Wireless Networks Anil Kumar S. Vullikanti Network Dynamics and Simulation Science

Capacity of Wireless Networks Anil Kumar S. Vullikanti Network Dynamics and Simulation Science Laboratory, Virginia Bioinformatics Institute and Department of Computer Science, Virginia Tech Anil Vullikanti (Virginia Tech) Capacity of Wireless

1.24k views • 108 slides
A Resource Allocation Controller for Cloud-based Adaptive Video Streaming Luca De Cicco , Saverio

A Resource Allocation Controller for Cloud-based Adaptive Video Streaming Luca De Cicco , Saverio

1/16 A Resource Allocation Controller for Cloud-based Adaptive Video Streaming Luca De Cicco , Saverio Mascolo, Dario Calamita Politecnico di Bari, Dipartimento di Ingegneria Elettrica e dellInformazione MCN 2013 - Budapest, Hungary 13 June

564 views • 27 slides
Security in Pervasive Wireless Security in Pervasive Wireless Systems Systems Wade Trappe

Security in Pervasive Wireless Security in Pervasive Wireless Systems Systems Wade Trappe

Security in Pervasive Wireless Security in Pervasive Wireless Systems Systems Wade Trappe Breaking Down the Issues (summary) Breaking Down the Issues (summary) Wireless is easy to sniff. We still need encryption services and key management.

533 views • 13 slides
Saliency and KAZE Features Authors: Siddharth Srivastava, Prerana Mukherjee, Dr. Brejesh Lall

Saliency and KAZE Features Authors: Siddharth Srivastava, Prerana Mukherjee, Dr. Brejesh Lall

Adaptive Image Compression Using Saliency and KAZE Features Authors: Siddharth Srivastava, Prerana Mukherjee, Dr. Brejesh Lall Department of Electrical Engineering Indian Institute of Technology, Delhi SPCOM 2016 Overview Introduction

326 views • 31 slides
Quadtrees: Hierarchical Grids Steve Oudot From S. Har-Peleds notes, Chapter 2 Outline

Quadtrees: Hierarchical Grids Steve Oudot From S. Har-Peleds notes, Chapter 2 Outline

CS468, Mon. Oct. 2nd, 2006 Quadtrees: Hierarchical Grids Steve Oudot From S. Har-Peleds notes, Chapter 2 Outline Examples and preliminary results. (deterministic) Static setting: compressed quadtrees. Dynamic setting:

1.18k views • 60 slides
Tradeoffs in XML Database Compression James Cheney University of Edinburgh Data Compression

Tradeoffs in XML Database Compression James Cheney University of Edinburgh Data Compression

Tradeoffs in XML Database Compression James Cheney University of Edinburgh Data Compression Conference March 30, 2006 Tradeoffs in XML Database Compression p.1/22 XML Compression XML: a format for tree-structured data Increasingly used

382 views • 22 slides
Elimination Service for Enterprises Ram Ramjee Microsoft Research India Bhavish Aggarwal^,

Elimination Service for Enterprises Ram Ramjee Microsoft Research India Bhavish Aggarwal^,

EndRE: An End-System Redundancy Elimination Service for Enterprises Ram Ramjee Microsoft Research India Bhavish Aggarwal^, Aditya Akella*, Ashok Anand*, Athula Balachandran~, Pushkar Chitnis^, Chitra Muthukrishnan*, and George Varghese# ^:

414 views • 24 slides

More recommend