c a n f p g a s c o m p e t e w i t h g p u s
play

C a n F P G A s c o m p e t e w i t h G P U s - PowerPoint PPT Presentation

Jul 09, 2023 •243 likes •599 views

G P U T e c h n o l o g y C o n f e r e n c e ( G T C 1 9 ) M a r c h 1 8 - 2 1 , 2 0 1 9 C a n F P G A s c o m p e t e w i t h G P U s ? J o h n R o m e i n , B r

  1. G P U T e c h n o l o g y C o n f e r e n c e ( G T C ’ 1 9 ) M a r c h 1 8 - 2 1 , 2 0 1 9 C a n F P G A s c o m p e t e w i t h G P U s ? J o h n R o m e i n , B r a m V e e n b o e r

  2. O u t l i n e ● e x p l a i n F P G A – h a r d w a r e ● F P G A v s . G P U – p r o g r a m m i n g m o d e l s ( O p e n C L ) – c a s e s t u d i e s ● m a t r i x m u l t i p l i c a t i o n a n a l y z e p e r f o r m a n c e & e n e r g y e fg i c i e n c y ● r a d i o - a s t r o n o m i c a l i m a g i n g – l e s s o n s l e a r n e d ● a n s w e r t h e q u e s t i o n i n t h e t i t l e 2

  3. W h a t i s a F i e l d - P r o g r a m m a b l e G a t e A r r a y ( F P G A ) ? ● c o n f i g u r a b l e p r o c e s s o r ● c o l l e c t i o n o f – m u l t i p l i e r s / a d d e r s – r e g i s t e r s – m e m o r i e s – l o g i c – t r a n s c e i v e r s – c l o c k s – i n t e r c o n n e c t 3

  4. W h a t i s a n F P G A p r o g r a m ? ● c o n n e c t e l e m e n t s – p e r f o r m s f i x e d f u n c t i o n ● d a t a - f l o w e n g i n e ● H a r d w a r e D e s c r i p t i o n L a n g u a g e ( H D L ) – V e r i l o g , V H D L – d i fg i c u l t 4

  5. F P G A v s G P U • F • G P G A a d v a n t a g e s P U a d v a n t a g e s – h – e i g h e n e r g y e fg i c i e n c y a s i e r t o p r o g r a m • n – m o i n s t r u c t i o n d e c o d i n g e t c . o r e f l e x i b l e • u s e / m o v e a s f e w b i t s a s p o s s i b l e – s h o r t c o m p i l a t i o n t i m e s – c o n f i g u r a b l e I / O • h i g h b a n d w i d t h • e . g . , m a n y t i m e s 1 0 0 G b E 5

  6. N e w I n t e l ( A l t e r a ) F P G A t e c h n o l o g i e s 1 )h i g h - l e v e l p r o g r a m m i n g l a n g u a g e ( O p e n C L ) 2 )h a r d F l o a t i n g - P o i n t U n i t s 3 )t i g h t i n t e g r a t i o n w i t h C P U c o r e s ➔ s i m p l e t a s k s : l e s s p r o g r a m m i n g e fg o r t t h a n H D L ➔ a l l o w s c o m p l e x H P C a p p l i c a t i o n s 6

  7. F P G A ≠ G P U • h a r d w a r e – G P U : u s e h a r d w a r e – F P G A : c r e a t e h a r d w a r e • e x e c u t i o n m o d e l – G P U : i n s t r u c t i o n s – F P G A : d a t a f l o w 7

  8. C o m m o n l a n g u a g e : O p e n C L • O p e n C L – s i m i l a r t o C U D A – C + e x p l i c i t p a r a l l e l i s m + s y n c h r o n i z a t i o n + s o fu w a r e - m a n a g e d c a c h e – o fg l o a d t o G P U / F P G A • G P U p r o g r a m s n o t s u i t a b l e f o r F P G A s – F P G A : d a t a - f l o w e n g i n e 8

  9. F P G A : d a t a - f l o w p i p e l i n e e x a m p l e C A B A B C r r r i i i • c r e a t e h a r d w a r e f o r c o m p l e x m u l t i p l y - a d d C.real += A.real * B.real; C.real += -A.imag * B.imag; F P U F P U C.imag += A.real * B.imag; C.imag += A.imag * B.real; • n e e d s f o u r F P U s • n e w i n p u t e n t e r s e v e r y c y c l e F P U F P U C C r i 9

  10. L o c a l m e m o r y • G P U : u s e m e m o r y , F P G A : c r e a t e m e m o r y f l o a t t m p [ 1 2 8 ] _ _ a t t r i b u t e _ _ ( ( …) ) ; • p r o p e r t i e s : – r e g i s t e r s o r m e m o r y b l o c k s – # b a n k s – b a n k w i d t h – b a n k a d d r e s s s e l e c t i o n b i t s a t t r i b u t e s – # r e a d p o r t s – # w r i t e p o r t s – s i n g l e / d o u b l e p u m p e d – w i t h / w i t h o u t a r b i t e r – r e p l i c a t i o n f a c t o r • w e l l d e s i g n e d p r o g r a m : f e w r e s o u r c e s , s t a l l - f r e e 1 0

  11. R u n n i n g m u l t i p l e k e r n e l s G P U : k e r n e l _ 1 d e v i c e me mo r y k e r n e l _ 2 F P G A : k e r n e l _ 1 c h a n n e l k e r n e l _ 2 • G P U : c o n s e c u t i v e l y • F P G A : c o n c u r r e n t l y – c h a n n e l s ( O p e n C L e x t e n s i o n ) c h a n n e l f l o a t 2 m y _ c h a n n e l _ _ a t t r i b u t e _ _ ( ( d e p t h ( 2 5 6 ) ) ) ; – r e q u i r e s l e s s m e m o r y b a n d w i d t h – a l l k e r n e l s r e s i d e n t → m u s t f i t 1 1

  12. P a r a l l e l c o n s t r u c t s i n O p e n C L C A B A B C r r r i i i • w o r k g r o u p s , w o r k i t e m s ( C U D A : t h r e a d b l o c k s , t h r e a d s ) – G P U : p a r a l l e l F P U F P U – F P G A : d e f a u l t : o n e w o r k i t e m / c y c l e ; n o t u s e f u l • F P G A : – c o m p i l e r a u t o - p a r a l l e l i z e s w h o l e k e r n e l F P U F P U – # p r a g m a u n r o l l • c r e a t e h a r d w a r e f o r c o n c u r r e n t l o o p i t e r a t i o n s C C r i – r e p l i c a t e k e r n e l s # p r a g ma u n r o l l f o r ( i n t i = 0 ; i < 3 ; i + + ) a [ i ] = i ; 0 1 2 1 2

  13. M o r e G P U / F P G A d i fg e r e n c e s • c o d e o u t s i d e c r i t i c a l l o o p : – G init_once(); P U : n o t a n i s s u e for (int i = 0; i < 100000; i ++) – F P G A : c a n u s e m a n y r e s o u r c e s do_many_times(); • S h i fu r e g i s t e r s for (int i = 256; i > 0; i --) a[i] = a[i – 1]; – F P G A : s i n g l e c y c l e – G P U : e x p e n s i v e 1 3

  14. R e s o u r c e o p t i m i z a t i o n s • c o m p i l e r f e e d b a c k – a fu e r m i n u t e s , n o t h o u r s – H T M L - b a s e d r e p o r t s • i n d i s p e n s i b l e t o o l 1 4

  15. F r e q u e n c y o p t i m i z a t i o n ● C o m p i l e r ( p l a c e & r o u t e ) d e t e r m i n e s F m a x – U n l i k e G P U s ● F u l l F P G A : l o n g e s t p a t h l i m i t s F m a x ● H D L : f i n e - g r a i n e d c o n t r o l ● O p e n C L : o n e c l o c k f o r f u l l d e s i g n ● F P G A : 4 5 0 M H z ; B S P : 4 0 0 M h z ; 1 D S P : 3 5 0 M h z ● L i t t l e c o m p i l e r f e e d b a c k ● R e c o m p i l e w i t h r a n d o m s e e d s ● C o m p i l e k e r n e l s i n i s o l a t i o n t o f i n d f r e q u e n c y l i m i t e r s – T e d i o u s , b u t u s e f u l 1 5

  16. A p p l i c a t i o n s ● d e n s e m a t r i x m u l t i p l i c a t i o n ● r a d i o - a s t r o n o m i c a l i m a g e r ● S i g n a l p r o c e s s i n g ( F I R f i l t e r , F F T , c o r r e l a t i o n s , b e a m f o r m i n g ) 1 6

  17. D e n s e m a t r i x m u l t i p l i c a t i o n • c o m p l e x < f l o a t > • h o r i z o n t a l & v e r t i c a l m e m o r y a c c e s s e s ( → c o a l e s c e ) • r e u s e o f d a t a ( → c a c h e ) 1 7

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

INTRODUCTION OF AN ORDINANCE PROVIDING FOR A TEMPORARY REDUCTION OF THE MARIJUANA BUSINESS

INTRODUCTION OF AN ORDINANCE PROVIDING FOR A TEMPORARY REDUCTION OF THE MARIJUANA BUSINESS

INTRODUCTION OF AN ORDINANCE PROVIDING FOR A TEMPORARY REDUCTION OF THE MARIJUANA BUSINESS TAX March 3, 2020 2 OVERVIEW Introduce the first reading of an ordinance providing for a 24-month temporary reduction of the business tax

248 views • 23 slides
Purpose Share with the Board our insights relating to new trends in customer growth and

Purpose Share with the Board our insights relating to new trends in customer growth and

Electrical Load Growth - Quarterly Update 1/8/2018 1 Chad Rissman- Director of Engineering and Asset Management &amp; Andy Wendell Director of Customer Services Purpose Share with the Board our insights relating to new trends in customer

330 views • 10 slides
Low Complexity Real-Time Simultaneous Localization and Mapping Using Velodyne LiDAR Sensor Dr.

Low Complexity Real-Time Simultaneous Localization and Mapping Using Velodyne LiDAR Sensor Dr.

Low Complexity Real-Time Simultaneous Localization and Mapping Using Velodyne LiDAR Sensor Dr. Kiran Gunnam/Algorithms Group www.velodynelidar.com Director of Algorithms, Velodyne LiDAR, Inc. CONFIDENTIAL GTC-2018, March 29 th , 2018. Outline

461 views • 23 slides
TDM, and Wireless with MAPS 818 West Diamond Avenue - Third Floor, Gaithersburg, MD 20878

TDM, and Wireless with MAPS 818 West Diamond Avenue - Third Floor, Gaithersburg, MD 20878

Traffic Simulation over IP, Ethernet, TDM, and Wireless with MAPS 818 West Diamond Avenue - Third Floor, Gaithersburg, MD 20878 Phone: (301) 670-4784 Fax: (301) 670-9187 Email: info@gl.com Website: http://www.gl.com 1 Traffic Simulation

1.17k views • 83 slides
Science Management Partnerships at the Landscape Scale: h d l Learning From and Partnering

Science Management Partnerships at the Landscape Scale: h d l Learning From and Partnering

Science Management Partnerships at the Landscape Scale: h d l Learning From and Partnering With g g the USDA Forest Service March 28, 2012 National Landscape Conservation N i l L d C i Cooperative Workshop Science Management

859 views • 82 slides
Innovative Strategies for Skyrocketing Healthcare Costs Tonya M allory U.S. Chamber of Commerce-

Innovative Strategies for Skyrocketing Healthcare Costs Tonya M allory U.S. Chamber of Commerce-

Innovative Strategies for Skyrocketing Healthcare Costs Tonya M allory U.S. Chamber of Commerce- Innovations in Workplace and Community Wellness: Navigating the New Terrain April 7, 2014 Agenda Trends Healthcare is moving toward

554 views • 30 slides
U SING A C USTOM -B UILT HDL FOR P RINTED C IRCUIT B OARD D ESIGN C APTURE Brent Nelson Brad

U SING A C USTOM -B UILT HDL FOR P RINTED C IRCUIT B OARD D ESIGN C APTURE Brent Nelson Brad

2.1 2.1 U SING A C USTOM -B UILT HDL FOR P RINTED C IRCUIT B OARD D ESIGN C APTURE Brent Nelson Brad Riching Josh Mangelson Dept. of Electrical and Computer Engineering Brigham Young University PCB West Presentation September 2012

1.1k views • 78 slides
Fault Injection in Mixed-Signal Environment Using Behavioral Fault Modeling in Verilog-A Seyed

Fault Injection in Mixed-Signal Environment Using Behavioral Fault Modeling in Verilog-A Seyed

Sharif University of Technology Department of Computer Engineering Dependable System Lab [DSL] Fault Injection in Mixed-Signal Environment Using Behavioral Fault Modeling in Verilog-A Seyed Nematollah Ahmadian, Seyed Ghassem Miremadi

398 views • 17 slides
An i-PARIHS evaluation of implementation fidelity for a worksite sedentary behavior intervention

An i-PARIHS evaluation of implementation fidelity for a worksite sedentary behavior intervention

An i-PARIHS evaluation of implementation fidelity for a worksite sedentary behavior intervention Matthew Buman, PhD Co-Investigators: Sarah Mullane*, Sarah Rydell, Meynard Toledo, Paul Estabrooks, Mark Pereira R01CA198971 Physical actvity AND

500 views • 23 slides
Exploring the Utility of Health Information Network Data as a Surveillance/Screening Tool in

Exploring the Utility of Health Information Network Data as a Surveillance/Screening Tool in

Exploring the Utility of Health Information Network Data as a Surveillance/Screening Tool in Addressing Health Implications of Environmental Exposures NAHDO 2019 Health Care Data Summit November 6-8, 2019 Little Rock, Arkansas Suzanne K.

322 views • 5 slides
High-Level and Model-Based Design Targeting FPGAs and SoCs Sander Ter Burg, FPGA System Engineer

High-Level and Model-Based Design Targeting FPGAs and SoCs Sander Ter Burg, FPGA System Engineer

CO-DEVELOPMENT MANUFACTURING INNOVATION &amp; SUPPORT High-Level and Model-Based Design Targeting FPGAs and SoCs Sander Ter Burg, FPGA System Engineer 3T B.V. What we do: Electronic and Embedded Systems Co-Development and Re-design

516 views • 27 slides
The thesis presentation will be a 3 slide wide setup containing the project outline throughout the

The thesis presentation will be a 3 slide wide setup containing the project outline throughout the

The thesis presentation will be a 3 slide wide setup containing the project outline throughout the entire presentation on the left slide, text/tables on the middle slide and pictures on the right slide. I. Project Overview a. Presentation Outline

239 views • 3 slides
Health Policy Commission Board Meeting CMIR Presentation July 18, 2018 Overview of Cost and

Health Policy Commission Board Meeting CMIR Presentation July 18, 2018 Overview of Cost and

Health Policy Commission Board Meeting CMIR Presentation July 18, 2018 Overview of Cost and Market Impact Reviews Market structure and new provider changes, including consolidations and alignments, have been shown to impact health care

828 views • 64 slides
AMN Healthcare Investor Presentation December 2018 The Innovator in Healthcare Workforce

AMN Healthcare Investor Presentation December 2018 The Innovator in Healthcare Workforce

AMN Healthcare Investor Presentation December 2018 The Innovator in Healthcare Workforce Solutions and Staffing Services Forward-Looking Statements This investor presentation contains forward- The Companys actual results could looking

450 views • 32 slides
Federal Street Acquisition Corp. and Agiliti Important information The information in this

Federal Street Acquisition Corp. and Agiliti Important information The information in this

Federal Street Acquisition Corp. and Agiliti Important information The information in this presentation relates to a proposed business combination among Federal Street Acquisition Corp. (FSAC) and Universal Hospital Services, Inc.

651 views • 28 slides
Community Care Collaborative Fiscal Year 2017 Proposed Budget C E N T R A L H E A L T H B O A R

Community Care Collaborative Fiscal Year 2017 Proposed Budget C E N T R A L H E A L T H B O A R

Community Care Collaborative Fiscal Year 2017 Proposed Budget C E N T R A L H E A L T H B O A R D O F M A N A G E R S J U L Y 2 7 , 2 0 1 6 P A T R I C I A Y O U N G B R O W N , C C C B O A R D C H A I R S A R A H C O O K , C C C

632 views • 27 slides
in Kansas and Missouri Results from the 2017 Kansas and Missouri Consumer Health Access Survey

in Kansas and Missouri Results from the 2017 Kansas and Missouri Consumer Health Access Survey

A View of Consumer Health Access in Kansas and Missouri Results from the 2017 Kansas and Missouri Consumer Health Access Survey (KMHS) KMHS Funders Health Care Foundation of Greater Kansas City Kansas Health Foundation Missouri Foundation for

919 views • 63 slides
Barclays Global Healthcare Conference Investor Presentation March 2019 Forward-looking

Barclays Global Healthcare Conference Investor Presentation March 2019 Forward-looking

Barclays Global Healthcare Conference Investor Presentation March 2019 Forward-looking Statements and Non-GAAP Information This presentation may include projections and other forward -looking statements within the meaning of the

436 views • 17 slides
Health Care Cost Trends Steering Committee FIRST MEETING, AUGUST 29, 2018 Agenda 1. Welcome

Health Care Cost Trends Steering Committee FIRST MEETING, AUGUST 29, 2018 Agenda 1. Welcome

Rhode Island Health Care Cost Trends Steering Committee FIRST MEETING, AUGUST 29, 2018 Agenda 1. Welcome and Introductions 9:00 am 9:15 am 2. Cost Trends Project Overview 9:15 am 9:40 am 3. Cost Trends Project Context 9:40 am

1.18k views • 72 slides
Estate Planning For The 21st Century Presenters Name Event Date IMPORTANT SLIDE INFORMATION

Estate Planning For The 21st Century Presenters Name Event Date IMPORTANT SLIDE INFORMATION

Estate Planning For The 21st Century Presenters Name Event Date IMPORTANT SLIDE INFORMATION NAELA Members The following slides were graciously provided by NAELA Member Rajiv Nagaich from Washington state. NAELA members have

615 views • 25 slides
Patient Advocates Program Teach-In 1/23/18 Protecting Rights of Older Persons When They Have

Patient Advocates Program Teach-In 1/23/18 Protecting Rights of Older Persons When They Have

Patient Advocates Program Teach-In 1/23/18 Protecting Rights of Older Persons When They Have Delegated Some of Their Health Care Decision Making Thomas Caprio, MD, University of Rochester Debby Rosenfeld, Esq., Ronald Fatoullah &amp; Associates

307 views • 5 slides
Trusted Health Plan This program is funded in part by the Government of the District of Columbia

Trusted Health Plan This program is funded in part by the Government of the District of Columbia

Case Management &amp; Care Coordination Trusted Health Plan This program is funded in part by the Government of the District of Columbia Department of Health Care Finance Trusted Health Plan Case Management Program Trusteds Case Management

526 views • 24 slides
VIATRIS: VIATRIS: TM TM A New Champion for Global Health A New Champion for Global Health

VIATRIS: VIATRIS: TM TM A New Champion for Global Health A New Champion for Global Health

VIATRIS: VIATRIS: TM TM A New Champion for Global Health A New Champion for Global Health February 27, 2020 1 Forward-Looking Statements This presentation contains forward-looking statements. Such forward-looking statements may

534 views • 7 slides
Barbara Bonnie W. Sonneborn, Esq. March 21, 2017 1) To understand the legal requirements of

Barbara Bonnie W. Sonneborn, Esq. March 21, 2017 1) To understand the legal requirements of

Barbara Bonnie W. Sonneborn, Esq. March 21, 2017 1) To understand the legal requirements of the Baker Act and its proper application in a hospital setting (both ER and inpatient) in order to meet the standard of care for physicians and the

1.54k views • 109 slides

More recommend