approximate program synthesis
play

Approximate Program Synthesis James Bornholt Emina Torlak Luis - PowerPoint PPT Presentation

May 20, 2023 •382 likes •775 views

Approximate Program Synthesis James Bornholt Emina Torlak Luis Ceze Dan Grossman University of Washington Writing approximate programs is hard Precise Implementation Writing approximate programs is hard Precise Implementation Writing

  1. Approximate Program Synthesis James Bornholt Emina Torlak Luis Ceze Dan Grossman University of Washington

  2. Writing approximate programs is hard Precise Implementation

  3. Writing approximate programs is hard Precise Implementation

  4. Writing approximate programs is hard Precise Implementation Desired Quality

  5. Writing approximate programs is hard Precise Implementation Approximate Compiler ACCEPT, Chisel, iACT, ExpAX, … Desired Quality

  6. Writing approximate programs is hard Precise Implementation Approximate Approximate Compiler Program ACCEPT, Chisel, iACT, ExpAX, … Desired Quality

  7. Writing approximate programs is hard s p e c 
 i fj c 
 m a p a p r g r o o x r i p m a t 
 i o n s n o o i p t a p t o o r n t u n n a i t i e s Precise Implementation Approximate Approximate Compiler Program ACCEPT, Chisel, iACT, ExpAX, … target 
 Desired Quality language q u global 
 a 
 l i t y m e t optimization r i c

  8. Synthesis: write programs automatically Programs Semantics

  9. Synthesis: write programs automatically Target Behavior f ( x ) = 4 x + 1 Programs Semantics

  10. Synthesis: write programs automatically Target Behavior f ( x ) = 4 x + 1 Programs Semantics

  11. Synthesis: write programs automatically 4*x + 1 Target Behavior f ( x ) = 4 x + 1 Programs Semantics

  12. Synthesis: write programs automatically 4*x + 1 Target Behavior f ( x ) = 4 x + 1 Programs Semantics

  13. Synthesis: write programs automatically 4*x + 1 (x<<2) + 1 Target Behavior f ( x ) = 4 x + 1 x + x + x + x + 1 Programs Semantics

  14. Synthesizing approximate programs 4*x + 1 (x<<2) + 1 Target Behavior f ( x ) = 4 x + 1 x + x + x + x + 1 Programs Semantics

  15. Synthesizing approximate programs 4*x + 1 Approximate (x<<2) + 1 Behavior Target Behavior f ( x ) = 4 x + 1 x + x + x + x + 1 Programs Semantics

  16. Synthesizing approximate programs 4*x + 1 Approximate (x<<2) + 1 Behavior Target Behavior f ( x ) = 4 x + 1 x<<2 x + x + x + x + 1 Programs Semantics

  17. Synthesizing approximate programs 4*x + 1 Approximate κ = 10 (x<<2) + 1 Behavior κ = 6 Target Behavior = κ 4 f ( x ) = 4 x + 1 x<<2 8 = κ x + x + x + x + 1 Programs Semantics

  18. Synthesizing approximate programs 4*x + 1 Approximate κ = 10 (x<<2) + 1 Behavior κ = 6 Target Behavior κ = 4 f ( x ) = 4 x + 1 x<<2 8 = κ x + x + x + x + 1 Programs Semantics

  19. Synthesis automates approximation s p e c 
 i fj c 
 m a p a p r g r o o x r i p m a t 
 i o n s n o o i p t a p t o o r n t u n n a i t i e s Precise Implementation Approximate Approximate Compiler Program ACCEPT, Chisel, iACT, ExpAX, … target 
 Desired Quality language q u global 
 a 
 l i t y m e t optimization r i c

  20. Synthesis automates approximation s p e c 
 i fj c 
 m a p a p r g r o o x r i p m a t 
 i o n s n o o i p t a p t o o r n t u n n a i t i e s Precise Implementation Approximate Approximate Program Program Synthesis target 
 Desired Quality language q u global 
 a 
 l i t y m e t optimization r i c

  21. Synthesis automates approximation speci fj c 
 Precise approximation 
 Implementation opportunities Approximate Approximate Program Program Synthesis target 
 Desired Quality global 
 program 
 language annotations optimization q u a 
 l i t y m e t r i c

  22. Existing synthesizers don’t scale enough Approximate benchmarks • fft • kmeans • inversek2j • sobel ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  23. Existing synthesizers don’t scale enough Approximate benchmarks • fft • kmeans • inversek2j • sobel Off-the-shelf synthesizers † • Symbolic ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% • Stochastic • Brute-force † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  24. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { Approximate benchmarks float r = 0; • fft r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); • kmeans r += (p[2] - c[2])*(p[2] - c[2]); • inversek2j float ret = sqrt(r); • sobel return ret; } float f(float p[3], float c[3]) { ?? } Off-the-shelf synthesizers † • Symbolic assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% • Stochastic • Brute-force † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  25. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { Approximate benchmarks float r = 0; • fft r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); • kmeans r += (p[2] - c[2])*(p[2] - c[2]); • inversek2j float ret = sqrt(r); • sobel return ret; } float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … Off-the-shelf synthesizers † • Symbolic assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% • Stochastic • Brute-force † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  26. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { Approximate benchmarks float r = 0; • fft r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); • kmeans r += (p[2] - c[2])*(p[2] - c[2]); • inversek2j float ret = sqrt(r); • sobel return ret; } float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … Off-the-shelf synthesizers † • Symbolic assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% • Stochastic • Brute-force Reference Program being program synthesized † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  27. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { Approximate benchmarks float r = 0; • fft r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); • kmeans r += (p[2] - c[2])*(p[2] - c[2]); • inversek2j float ret = sqrt(r); • sobel return ret; } float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … Off-the-shelf synthesizers † • Symbolic assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% • Stochastic • Brute-force Reference Program being program synthesized † Alur et al. Syntax-Guided Synthesis . FMCAD 2013.

  28. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); return ret; } float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

  29. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); return ret; } float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

  30. Existing synthesizers don’t scale enough float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); 7.1 × 10 43 return ret; } programs float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

  31. Use reference programs to guide synthesis float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); 7.1 × 10 43 return ret; } programs float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

  32. Use reference programs to guide synthesis float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); 7.1 × 10 43 return ret; } programs float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

  33. Use reference programs to guide synthesis float dist(float p[3], float c[3]) { float r = 0; r += (p[0] - c[0])*(p[0] - c[0]); r += (p[1] - c[1])*(p[1] - c[1]); r += (p[2] - c[2])*(p[2] - c[2]); float ret = sqrt(r); 7.1 × 10 43 return ret; } programs float f(float p[3], float c[3]) { ?? + - * / & | ^ } << >> … assert ∀ p, c. | dist ( p, c ) − f ( p, c ) | < 50% Programs Reference Program being program synthesized

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

From Program Synthesis to Optimal Program . . . Optimal Program Synthesis Logical Interpretation

From Program Synthesis to Optimal Program . . . Optimal Program Synthesis Logical Interpretation

Need for Data . . . Program Synthesis Wave Algorithm for . . . Boolean Logic . . . From Program Synthesis to Optimal Program . . . Optimal Program Synthesis Logical Interpretation . . . A Seemingly Natural . . . Counter-Example to . . .

725 views • 24 slides
Synthesis of Ranking Functions and Synthesis of Inductive Invariants and Synthesis of

Synthesis of Ranking Functions and Synthesis of Inductive Invariants and Synthesis of

Synthesis of Ranking Functions and Synthesis of Inductive Invariants and Synthesis of Recurrence Sets via Constraint Solving Andreas Podelski January 17, 2012 1 Program Verification and Constraints Reasoning about program

422 views • 28 slides
QBF-BASED SYNTHESIS OF OPTIMAL WORD-SPLITTING IN APPROXIMATE MULTI-LEVEL CELLS Daniel E. Holcomb

QBF-BASED SYNTHESIS OF OPTIMAL WORD-SPLITTING IN APPROXIMATE MULTI-LEVEL CELLS Daniel E. Holcomb

QBF-BASED SYNTHESIS OF OPTIMAL WORD-SPLITTING IN APPROXIMATE MULTI-LEVEL CELLS Daniel E. Holcomb Kevin Fu University of Michigan 2 March 2014 QBF Synthesis of Optimal Word-Splitting 2 Motivation 2 March 2014 QBF Synthesis of Optimal

752 views • 29 slides
Guarantees in in Program Synthesis Qinheping Hu , Jason Breck , John Cyphert , Loris D'Antoni ,

Guarantees in in Program Synthesis Qinheping Hu , Jason Breck , John Cyphert , Loris D'Antoni ,

Guarantees in in Program Synthesis Qinheping Hu , Jason Breck , John Cyphert , Loris D'Antoni , Thomas Reps Program Synthesis Specification Solution Program Synthesizer Search space Program Synthesis is Un Unpredicta edictable ble

1.05k views • 41 slides
From Program Verification to Program Synthesis Overview Jaak Ristioja March 30, 2010 1 / 91

From Program Verification to Program Synthesis Overview Jaak Ristioja March 30, 2010 1 / 91

From Program Verification to Program Synthesis Overview Jaak Ristioja March 30, 2010 1 / 91 Reference From Program Verification to Program Synthesis. @ POPL10; January 17-23, 2010 Saurabh Srivastava , University of Maryland, College Park

1.21k views • 91 slides
Conditional Program Generation for Bimodal Program Synthesis Swarat Chaudhuri Rice University

Conditional Program Generation for Bimodal Program Synthesis Swarat Chaudhuri Rice University

Conditional Program Generation for Bimodal Program Synthesis Swarat Chaudhuri Rice University www.cs.rice.edu/~swarat (Joint work with Chris Jermaine, Vijay Murali, and Letao Qi) Program synthesis [Simon 1963, Summers 1977, Manna-Waldinger

411 views • 22 slides
Recursive Program Synthesis Aws Albarghouthi (UToronto), Sumit Gulwani (MSR), and Zachary

Recursive Program Synthesis Aws Albarghouthi (UToronto), Sumit Gulwani (MSR), and Zachary

Recursive Program Synthesis Aws Albarghouthi (UToronto), Sumit Gulwani (MSR), and Zachary Kincaid (UToronto) CAV 2013 Saint Petersburg, Russia Program Synthesis Magic Specification Program 2 Program Synthesis Magic Specification

753 views • 74 slides
Syntax-Guided Synthesis Rajeev Alur University of Pennsylvania 1 Program Verification Program

Syntax-Guided Synthesis Rajeev Alur University of Pennsylvania 1 Program Verification Program

Syntax-Guided Synthesis Rajeev Alur University of Pennsylvania 1 Program Verification Program P Specification S Verifier Proof of correctness or Witness of a bug 2 Classical Program Synthesis Specification S High Level WHAT

1.28k views • 59 slides
Deep Learning over Big Code for Program Analysis and Synthesis Swarat Chaudhuri, Vijay

Deep Learning over Big Code for Program Analysis and Synthesis Swarat Chaudhuri, Vijay

Deep Learning over Big Code for Program Analysis and Synthesis Swarat Chaudhuri, Vijay Murali, Chris Jermaine Programming is hard Program synthesis, debugging, verification, repair Can we automate these processes? Decades of prior

1.59k views • 143 slides
Neural Program Synthesis with Priority Queue Training Daniel A. Abolafia, Mohammad Norouzi,

Neural Program Synthesis with Priority Queue Training Daniel A. Abolafia, Mohammad Norouzi,

Neural Program Synthesis with Priority Queue Training Daniel A. Abolafia, Mohammad Norouzi, Jonathan Shen, Rui Zhao, Quoc V. Le https://arxiv.org/abs/1801.03526 Why Program Synthesis? One of the hard AI reasoning domains A tool for

819 views • 27 slides
Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,)

Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,)

Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,) Armando)Solar@Lezama 2 ,)and)Jeffrey)S.)Foster 1) ) 1.)University)of)Maryland,)College)Park) 2.)MIT)CSAIL ) Syntax@guided)Synthesis) bit[32]

579 views • 57 slides
Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,)

Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,)

Adaptive)Concretization)for) Parallel)Program)Synthesis) Jinseong(Jeon 1 ,)Xiaokang)Qiu 2 ,) Armando)Solar@Lezama 2 ,)and)Jeffrey)S.)Foster 1) ) 1.)University)of)Maryland,)College)Park) 2.)MIT)CSAIL ) Syntax@guided)Synthesis) bit[32]

310 views • 27 slides
Program Synthesis Tikhon Jelvis (tikhon@jelv.is) February 24, 2014 Synthesis Find a program

Program Synthesis Tikhon Jelvis (tikhon@jelv.is) February 24, 2014 Synthesis Find a program

Program Synthesis Tikhon Jelvis (tikhon@jelv.is) February 24, 2014 Synthesis Find a program to some specification ( ( input , output ) ) input/output pairs executable specification P x . ( x , P ( x )) Why? easy

594 views • 23 slides
Improving Neural Program Synthesis with Inferred Execution Traces Richard Shin 1 Illia Polosukhin

Improving Neural Program Synthesis with Inferred Execution Traces Richard Shin 1 Illia Polosukhin

Improving Neural Program Synthesis with Inferred Execution Traces Richard Shin 1 Illia Polosukhin 2 Dawn Song 1 1 UC Berkeley 2 NEAR Protocol Poster: Room 210 &amp; 230 AB #31 Background For program synthesis from input-output examples ,

393 views • 16 slides
M emory A ugmented P olicy O ptimization ( MAPO ) for Program Synthesis and Semantic Parsing Chen

M emory A ugmented P olicy O ptimization ( MAPO ) for Program Synthesis and Semantic Parsing Chen

M emory A ugmented P olicy O ptimization ( MAPO ) for Program Synthesis and Semantic Parsing Chen Liang, Mohammad Norouzi, Jonathan Berant, Quoc Le, Ni Lao Program Synthesis / Semantic Parsing how many more passengers flew to los angeles than to

686 views • 29 slides
Program Boosting: Program Synthesis via Crowd-Sourcing Robert A. Cochran, Loris DAntoni,

Program Boosting: Program Synthesis via Crowd-Sourcing Robert A. Cochran, Loris DAntoni,

Program Boosting: Program Synthesis via Crowd-Sourcing Robert A. Cochran, Loris DAntoni, Benjamin Livshits, David Molnar and Margus Veanes Presented by: Sam Witty, Shehzaad Dhuliawala and Samer Nashed Outline Introduction (Research

734 views • 62 slides
COMP80122 Slides and Presentations with special thanks to Sebastian Brandt , to whom I owe most

COMP80122 Slides and Presentations with special thanks to Sebastian Brandt , to whom I owe most

COMP80122 Slides and Presentations with special thanks to Sebastian Brandt , to whom I owe most ideas for these slides: http://lat.inf.tu-dresden.de/~brandt/HowtoDesignaSlideshow.pdf Carole Goble | Uli Sattler School of Computer Science

675 views • 33 slides
LARGE CACHE DESIGN Mahdi Nazm Bojnordi Assistant Professor School of Computing University of

LARGE CACHE DESIGN Mahdi Nazm Bojnordi Assistant Professor School of Computing University of

LARGE CACHE DESIGN Mahdi Nazm Bojnordi Assistant Professor School of Computing University of Utah CS/ECE 7810: Advanced Computer Architecture Overview Upcoming deadline Feb. 3 rd : project group formation This lecture Gated Vdd/

467 views • 36 slides
2019 SECOND QUARTER EARNINGS CONFERENCE CALL August 6, 2019 Forward-looking Statement Except

2019 SECOND QUARTER EARNINGS CONFERENCE CALL August 6, 2019 Forward-looking Statement Except

2019 SECOND QUARTER EARNINGS CONFERENCE CALL August 6, 2019 Forward-looking Statement Except for historical information, all other information provided in this presentation consists of forward - looking statements within the meaning of the

721 views • 29 slides
T a i l r e c u r s i o n ( n o t o n e x a m ) H o w i s r e c u

T a i l r e c u r s i o n ( n o t o n e x a m ) H o w i s r e c u

T a i l r e c u r s i o n ( n o t o n e x a m ) H o w i s r e c u r s i o n i m p l e m e n t e d ? Wh e n y o u c a l l f u n c t i o n B f r o m f u n c t i o n A ,

954 views • 44 slides
Taking Part-Time Programmers Seriously Jesse A. Tov Elizabeth Tov Northeastern University

Taking Part-Time Programmers Seriously Jesse A. Tov Elizabeth Tov Northeastern University

Taking Part-Time Programmers Seriously Jesse A. Tov Elizabeth Tov Northeastern University Boston College OBT 2012 need your help. Part-time programmers Part-time programmers need your help. languages and tools Terrible tools No real

404 views • 16 slides
Why Batch and User Evaluations Do Not Give the Same Results A. Turpin Curtin University of

Why Batch and User Evaluations Do Not Give the Same Results A. Turpin Curtin University of

Why Batch and User Evaluations Do Not Give the Same Results A. Turpin Curtin University of Technology Perth, Australia W. Hersh Oregon Health Sciences University Portland, Oregon Presented at SIGIR2001 New Orleans TREC ad-hoc 0.4 MAP 0.3

607 views • 18 slides
Distributed goodness-of-fit: when you can't talk much and have little in common Jayadev Acharya

Distributed goodness-of-fit: when you can't talk much and have little in common Jayadev Acharya

Distributed goodness-of-fit: when you can't talk much and have little in common Jayadev Acharya (Cornell), Clment Canonne (Stanford), Yanjun Han (Stanford), Ziteng Sun (Cornell), and Himanshu Tyagi (IISc Bangalore)

219 views • 19 slides
Accelerator Driven Subcritical Reactors Introducing GEM*STAR A Particularly Advantageous

Accelerator Driven Subcritical Reactors Introducing GEM*STAR A Particularly Advantageous

Accelerator Driven Subcritical Reactors Introducing GEM*STAR A Particularly Advantageous Example Tom Roberts Muons, Inc. Technology to revitalize the nuclear power industry through improved safety, waste management, efficiency, and

795 views • 21 slides

More recommend