resource allocation for hardware implementations of map
play

Resource Allocation for Hardware Implementations of Map Richard - PowerPoint PPT Presentation

Jan 14, 2023 •249 likes •779 views

Resource Allocation for Hardware Implementations of Map Richard Townsend Martha A. Kim Stephen A. Edwards Columbia University ASBD Workshop, June 15, 2014 Functional Programs to Functional Hardware Functional program (Haskell) Compiler

  1. Resource Allocation for Hardware Implementations of Map Richard Townsend Martha A. Kim Stephen A. Edwards Columbia University ASBD Workshop, June 15, 2014

  2. Functional Programs to Functional Hardware Functional program (Haskell) Compiler Circuit (Verilog)

  3. Functional Programs to Functional Hardware Map f [ x 1 , x 2 ,..., x n ] Ordered List This Talk ?

  4. Functional Programs to Functional Hardware Map f [ x 1 , x 2 ,..., x n ] Ordered List This Talk Order Dependent ? scan fold

  5. Functional Map vs. MapReduce

  6. Functional Map vs. MapReduce (0,3) (1,3) (2,3) (3,3) (0,2) (1,2) (2,2) (3,2) (0,1) (1,1) (2,1) (3,1) (0,0) (1,0) (2,0) (3,0)

  7. Functional Map vs. MapReduce Ordered (0,3) (1,3) (2,3) (3,3) (0,2) (1,2) (2,2) (3,2) (0,1) (1,1) (2,1) (3,1) (0,0) (1,0) (2,0) (3,0)

  8. Functional Map vs. MapReduce Ordered (0,3) (1,3) (2,3) (3,3) (0,2) (1,2) (2,2) (3,2) (0,1) (1,1) (2,1) (3,1) (0,0) (1,0) (2,0) (3,0) Unordered

  9. Structure of a Hardware Implementation f

  10. Structure of a Hardware Implementation f f f f f

  11. Structure of a Hardware Implementation f f f f f

  12. Structure of a Hardware Implementation f f f f f x 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8 x 9 x 10

  13. Structure of a Hardware Implementation f f x 7 x 6 x 5 x 4 x 3 x 2 x 1 f f f

  14. Structure of a Hardware Implementation x 1 f x 2 f x 10 x 9 x 8 x 7 x 6 x 3 f x 4 f x 5 f

  15. Structure of a Hardware Implementation f f ( x 1 ) x 2 f x 10 x 9 x 8 x 7 x 6 f f ( x 3 ) x 4 f x 5 f

  16. Structure of a Hardware Implementation x 6 f f ( x 1 ) x 2 f x 10 x 9 x 8 x 7 f f ( x 3 ) x 4 f x 5 f

  17. Structure of a Hardware Implementation x 6 f x 2 f x 10 x 9 x 8 x 7 f f ( x 3 ) f ( x 1 ) x 4 f x 5 f

  18. Structure of a Hardware Implementation x 6 f x 2 f x 10 x 9 x 8 x 7 f ( x 7 ) f f ( x 3 ) f ( x 1 ) x 4 f x 5 f

  19. Structure of a Hardware Implementation still processing x 6 f x 2 f x 10 x 9 x 8 x 7 f ( x 7 ) f f ( x 3 ) f ( x 1 ) x 4 f x 5 f

  20. Structure of a Hardware Implementation still processing x 6 f stuck in buffer x 2 f x 10 x 9 x 8 x 7 f ( x 7 ) f f ( x 3 ) f ( x 1 ) x 4 f x 5 f

  21. Structure of a Hardware Implementation still processing x 6 f stuck in buffer x 2 f x 10 x 9 x 8 x 7 f ( x 7 ) f f ( x 3 ) f ( x 1 ) x 4 f holding and stalling x 5 f

  22. Structure of a Hardware Implementation f f f f f

  23. Structure of a Hardware Implementation f f f f f More Functional Units More Buffers (Parallelism) (Utilization)

  24. Multiple Possible Configurations...Which to Choose? Area = 15

  25. Multiple Possible Configurations...Which to Choose? Buffers 50% size of func. unit Area = 15

  26. Multiple Possible Configurations...Which to Choose? f f f f f f f 15 Functional Units f f f f Area = 15 f f f f

  27. Multiple Possible Configurations...Which to Choose? f 28 Buffers Area = 15

  28. Multiple Possible Configurations...Which to Choose? Area = 15 f f 3 f f 5 f f f 24 × 1 2 = 12 f 20 × 1 2 = 10

  29. Workload Structure f f f

  30. Workload Structure f f f Best-case

  31. Workload Structure Time f f f f f Best-case f

  32. Workload Structure Time f f f f f Best-case f ? Average-case

  33. Workload Structure Time f f f f f Best-case f ? Average-case Worst-case

  34. Workload Structure Time f f f f f Best-case f ? Average-case f f Worst-case f

  35. Workload Structure Time f f f f f Best-case f ? Average-case f f Worst-case f

  36. Optimal Resource Allocation Simulator Results 100% 80% Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  37. Optimal Resource Allocation Simulator Results 100% 80% Maximizing Functional Units Completion Time 60% 2x speedup 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  38. Optimal Resource Allocation Simulator Results 100% 80% Maximizing Buffers 3x speedup Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  39. Optimal Resource Allocation f f f f f 100% 80% Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  40. Optimal Resource Allocation f f f f f 100% 80% Completion Time 60% Slower 40% Fewer Buffers 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  41. Why Are There Multiple Optima? 100% 80% Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30

  42. Why Are There Multiple Optima? 100% 80% 12 Completion Time 60% 40% 5 20% 7 0% 0 5 10 15 20 25 30 10

  43. Why Are There Multiple Optima? 100% 80% 12 11 Completion Time 60% 40% 5 6 20% 7 6 0% 0 5 10 15 20 25 30 10 11

  44. Performance Scales with Area 100% 90% 80% 70% Completion Time 60% 50% 40% 30% Minimum 20% 10% Ideal 0% 0 10 20 30 40 50 60 Total Area

  45. Performance Scales with Area 100% Completion Time 80% 60% 40% 20% 0% 0 10 20 30 40 50 60 Buffers Slots / Functional Unit 30 20 10 0 0 10 20 30 40 50 60 10 Functional Units 8 6 4 2 0 10 20 30 40 50 60 Total Area

  46. Performance Scales with Area 100% Completion Time 80% 60% f 40% 20% f 0% 0 10 20 30 40 50 60 Buffers Slots / Functional Unit 30 20 10 0 0 10 20 30 40 50 60 12 Functional Units 10 8 6 4 2 0 10 20 30 40 50 60 Total Area

  47. Conclusions 100% 80% Area trade-off is important... Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit

  48. Conclusions 100% 80% Area trade-off is important... Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit f f f ...and non-obvious f f ? f f f

  49. Conclusions 100% 80% Area trade-off is important... Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit f f f ...and non-obvious f f ? f f f f Model helps explore design space f f f f x 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8 x 9 x 10

  50. Conclusions 100% 80% Area trade-off is important... Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit f f f ...and non-obvious f f ? f f f f Model helps explore design space f f f f x 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8 x 9 x 10 Synthesize Efficient Hardware Implementation of Map

  51. Conclusions 100% 80% Area trade-off is important... Completion Time 60% 40% 20% 0% 0 5 10 15 20 25 30 50 100 150 200 250 Buffer Slots per Functional Unit f f f ...and non-obvious f f ? f f f f Model helps explore design space f f f f x 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8 x 9 x 10 Synthesize Efficient Hardware Implementation of Map 100% 90% 80% 70% Completion Time Map Fold Scan 60% 50% Enhance our abstraction 40% 30% 20% 10% 0% 0 10 20 30 40 50 60 Total Area

Recommend

Section 6.1: Resource Allocation Issues Chapter 6: Congestion Control and Resource Allocation

Section 6.1: Resource Allocation Issues Chapter 6: Congestion Control and Resource Allocation

Section 6.1: Resource Allocation Issues Chapter 6: Congestion Control and Resource Allocation CS/ECPE 5516: Comm. Network Prof. Abrams Spring 2000 CS/ECPE5516 1 2 CS/ECPE5516 Whats wrong with doing How to prevent traffic jams

311 views • 6 slides
Resource Allocation Task Force Resource Allocation Task Force Gigi Karmous Edwards Gigi

Resource Allocation Task Force Resource Allocation Task Force Gigi Karmous Edwards Gigi

Resource Allocation Task Force Resource Allocation Task Force Gigi Karmous Edwards Gigi ke@ncsu.edu Gigi_ke@ncsu.edu GLIF Winter Meeting, Hong Kong 2011 Purpose of R Resource Allocation Task Force All i T k F What does the user want

342 views • 7 slides
Resource Allocation and Deadlock Resource Allocation and Deadlock Handling Conditions for

Resource Allocation and Deadlock Resource Allocation and Deadlock Handling Conditions for

Resource Allocation and Deadlock Resource Allocation and Deadlock Handling Conditions for Deadlock [Coffman-etal 1971] 4 conditions must hold simultaneously for a deadlock to occur: deadlock to occur: Mutual exclusion: only one process at

520 views • 31 slides
Dominant Resource Fairness: Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei

Dominant Resource Fairness: Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei

Dominant Resource Fairness: Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei Zaharia, Benjamin Hindman, Andy Konwinski, Scott Shenker, Ion Stoica UC Berkeley Introduction: Resource Allocation There is a need to design fair,

604 views • 22 slides
Hardware- -Based Implementations Based Implementations Hardware of Factoring Algorithms of

Hardware- -Based Implementations Based Implementations Hardware of Factoring Algorithms of

Hardware- -Based Implementations Based Implementations Hardware of Factoring Algorithms of Factoring Algorithms Factoring Large Numbers with the TWIRL Device Factoring Large Numbers with the TWIRL Device Adi Shamir, Eran Tromer Adi

469 views • 42 slides
Adversarial Risk Analysis Models for Urban Security Resource Allocation Urban Security Resource

Adversarial Risk Analysis Models for Urban Security Resource Allocation Urban Security Resource

Adversarial Risk Analysis Models for Urban Security Resource Allocation Urban Security Resource Allocation David Ros Insua, Royal Academy Cesar Gil, U. Rey Juan Carlos Jess Ros, IBM Research YH COST Smart Cities Wshop Paris, September

507 views • 22 slides
THE STATE-OF-THE-ART OF HARDWARE IMPLEMENTATIONS OF ELLIPTIC CURVE CRYPTOGRAPHY Kimmo Jrvinen

THE STATE-OF-THE-ART OF HARDWARE IMPLEMENTATIONS OF ELLIPTIC CURVE CRYPTOGRAPHY Kimmo Jrvinen

THE STATE-OF-THE-ART OF HARDWARE IMPLEMENTATIONS OF ELLIPTIC CURVE CRYPTOGRAPHY Kimmo Jrvinen Department of Computer Science University of Helsinki kimmo.u.jarvinen@helsinki.fi ECRYPT-CSA Workshop on Hardware Benchmarking Bochum, Germany,

840 views • 58 slides
D Dominant Resource Fairness (DRF) i t R F i (DRF) Fair Allocation of Multiple Resource Types

D Dominant Resource Fairness (DRF) i t R F i (DRF) Fair Allocation of Multiple Resource Types

D Dominant Resource Fairness (DRF) i t R F i (DRF) Fair Allocation of Multiple Resource Types Ali Ghodsi , Matei Zaharia Benjamin Hindman, Andy Konwinski, B j i Hi d A d K i ki Scott Shenker, Ion Stoica University of California, Berkeley

781 views • 51 slides
ISA Implementations Partly in Run programs for one ISA on hardware with different ISA Techniques:

ISA Implementations Partly in Run programs for one ISA on hardware with different ISA Techniques:

ISA Implementations Partly in Run programs for one ISA on hardware with different ISA Techniques: Software Emulation OS software interprets instructions at run-time E.g., OS for PowerPC Macs had emulator for 68000 code Run-time

358 views • 5 slides
Cost-Effective Resource Allocation for Deploying 1 Vinay Setty vsetty@mpi-inf.mpg.de Pub/Sub

Cost-Effective Resource Allocation for Deploying 1 Vinay Setty vsetty@mpi-inf.mpg.de Pub/Sub

Cost-Effective Resource Allocation for Deploying 1 Vinay Setty vsetty@mpi-inf.mpg.de Pub/Sub on Cloud Published in ICDCS 2014 Roman Vitenberg Maarten van Steen Gunnar Kreitz Guido Urdaneta Cost-Effective Resource Allocation for Deploying

904 views • 41 slides
Efficiency and Fairness in Distributed Resource Allocation Ulle Endriss Institute for Logic,

Efficiency and Fairness in Distributed Resource Allocation Ulle Endriss Institute for Logic,

Distributed Resource Allocation TU Delft Agent Colloquium: Nov 2006 Efficiency and Fairness in Distributed Resource Allocation Ulle Endriss Institute for Logic, Language and Computation University of Amsterdam joint work with

405 views • 23 slides
CS 514: Computer Networks Lecture 8: Router-Assisted Resource Allocation Xiaowei Yang

CS 514: Computer Networks Lecture 8: Router-Assisted Resource Allocation Xiaowei Yang

CS 514: Computer Networks Lecture 8: Router-Assisted Resource Allocation Xiaowei Yang xwy@cs.duke.edu Review A fundamental question of networking: who gets to send at what speed? Design Space for resource allocation Router-based

740 views • 46 slides
Heterogeneity-Aware Resource Allocation and Scheduling in the Cloud Gunho Lee (UC Berkeley)

Heterogeneity-Aware Resource Allocation and Scheduling in the Cloud Gunho Lee (UC Berkeley)

Heterogeneity-Aware Resource Allocation and Scheduling in the Cloud Gunho Lee (UC Berkeley) Byung-Gon Chun (Yahoo! Research) Randy H. Katz (UC Berkeley) We have resources and jobs Resource Job/Task Allocate resources (slots) Allocation

755 views • 23 slides
Gender inequality of educational resource allocation within household in China the effects of

Gender inequality of educational resource allocation within household in China the effects of

Gender inequality of educational resource allocation within household in China the effects of resource-limitation and resource-dilution Yixiao LIU Institute for Population and Development Studies School of Public Policy and Administration

472 views • 21 slides
Chapter 6 Cloud Resource Management and Scheduling Contents Resource management and

Chapter 6 Cloud Resource Management and Scheduling Contents Resource management and

Chapter 6 Cloud Resource Management and Scheduling Contents Resource management and scheduling. Policies and mechanisms. Applications of control theory to cloud resource allocation. Stability of a two-level resource allocation

1.47k views • 39 slides
ADEPT Scalability Predictor in Support of Adaptive Resource Allocation IPDPS 2010 Outline

ADEPT Scalability Predictor in Support of Adaptive Resource Allocation IPDPS 2010 Outline

Arash Deshmeh, Jacob Machina, and Angela C. Sodan University of Windsor, Canada ADEPT Scalability Predictor in Support of Adaptive Resource Allocation IPDPS 2010 Outline Background: Adaptive Resource Allocation Related Work Downey

329 views • 22 slides
Resource allocation strategies for multicarrier radio systems Marco Moretti Information

Resource allocation strategies for multicarrier radio systems Marco Moretti Information

Resource allocation strategies for multicarrier radio systems Marco Moretti Information Engineering Department Universit di Pisa marco.moretti@iet.unipi.it Summary Introduction Convex optimization Single-cell resource allocation

1.12k views • 110 slides
Hardware Implementations of Fixed-Point Atan2 Florent de Dinechin Matei I stoan Universit

Hardware Implementations of Fixed-Point Atan2 Florent de Dinechin Matei I stoan Universit

Hardware Implementations of Fixed-Point Atan2 Hardware Implementations of Fixed-Point Atan2 Florent de Dinechin Matei I stoan Universit e de Lyon, INRIA, INSA-Lyon, CITI-Lab ARITH22 Florent de Dinechin, Matei I stoan Hardware

582 views • 44 slides
Dominant Resource Fairness : Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei

Dominant Resource Fairness : Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei

Dominant Resource Fairness : Fair Allocation of Multiple Resource Types Ali Ghodsi, Matei Zaharia, Benjamin Hindman, Andy Konwinski, Scott Shenker, Ion Stoica University of California, Berkeley Talk by: Yael Amsterdamer Introduction Why

800 views • 22 slides
Dynamic Resource Allocation for Database Servers Running on Virtual Storage Gokul Soundararajan,

Dynamic Resource Allocation for Database Servers Running on Virtual Storage Gokul Soundararajan,

Dynamic Resource Allocation for Database Servers Running on Virtual Storage Gokul Soundararajan, Daniel Lupei, Saeed Ghanbari, Adrian Daniel Popescu, Jin Chen, Cristiana Amza University of Toronto 1 Multi-tier Resource Allocation

922 views • 44 slides
Bricks and Tools for Secure Hardware Implementations Francesco Regazzoni Francesco Regazzoni 06

Bricks and Tools for Secure Hardware Implementations Francesco Regazzoni Francesco Regazzoni 06

Bricks and Tools for Secure Hardware Implementations Francesco Regazzoni Francesco Regazzoni 06 June 2014, ibenik, Croatia P. 1 Why Electronic Design Automation? Surely the purpose of science is to ease human hardship Galileo,

846 views • 82 slides
Crime Research in Geography Resource allocation: we helped design the police Basic Command Unit

Crime Research in Geography Resource allocation: we helped design the police Basic Command Unit

Crime Research in Geography Resource allocation: we helped design the police Basic Command Unit families for the national resource allocation formula in the 1990s. More recently been advising West Yorkshire police. Predictive policing:

415 views • 28 slides
Decentralized Resource Allocation Mechanisms in Networks Tudor Stoenescu Information Science

Decentralized Resource Allocation Mechanisms in Networks Tudor Stoenescu Information Science

Decentralized Resource Allocation Mechanisms in Networks Tudor Stoenescu Information Science and Technology Caltech Organization of the Talk Major issues of resource allocation in networks Overview of fundamental issues in

1.56k views • 89 slides
Criteria for resource allocation ISPC Workshop 9 May 2017 Amsterdam http://ispc.cgiar.org

Criteria for resource allocation ISPC Workshop 9 May 2017 Amsterdam http://ispc.cgiar.org

Criteria for resource allocation ISPC Workshop 9 May 2017 Amsterdam http://ispc.cgiar.org Workshop on Resource Allocation Structure What we did to assess the quality of Phase II CRP proposals Update on work with the System to

449 views • 14 slides

More recommend