EnerJ

"Approximate Data Types for Safe and General Low-Power Computation"
EnerJ is a programming language extension enabling safe approximate computation. The paper was in PLDI 2011.

Transcript

1. EnerJ Approximate Data Types for Safe and General Low-Power Computation Adrian Sampson Werner Dietl Emily Fortuna Danushen Gnanapragasam Luis Ceze Dan Grossman sa pa PLDI 2011 University of Washington
2. photo: Apple, Inc.
3. photo: Robert Scoble (CC BY 2.0)
4. photo: Rob Van De Pavert (CC BY-NC-SA 2.0)
5. EnerJ: Save energy using programmer controls over execution correctness.
7. Perfect correctness is not required information retrieval machine learning sensory data scientific computing physical simulation games augmented reality computer vision
8. Anant Agarwal, Martin Rinard, Stelios Sidiroglou, Sasa Misailovic, and Henry Hoffmann. Using code perforation to improve performance, reduce energy consumption, and respond to failures. Technical report, MIT, 2009. [1] B.E.S. Akgul, L.N. Chakrapani, P. Korkmaz, and K.V. Palem. Probabilistic CMOS technology: A survey and future directions. In IFIP Intl. Conference on VLSI, 2006. [2] M. de Kruijf and K. Sankaralingam. Exploring the synergy of emerging workloads and silicon reliability trends. In SELSE, 2009. [3] Larkhoon Leem, Hyungmin Cho, Jason Bau, Quinn A. Jacobson, and Subhasish Mitra. ERSA: Error resilient system architecture for probabilistic applications. In DATE, 2010. [4] Xuanhua Li and Donald Yeung. Exploiting soft computing for increased fault tolerance. In ASGI, 2006. [5] Song Liu, Karthik Pattabiraman, Thomas Moscibroda, and Benjamin G. Zorn. Flicker: Saving refresh-power in mobile devices through critical data partitioning. Technical Report MSR-TR-2009-138, Microsoft Research, 2009. [6] Sriram Narayanan, John Sartori, Rakesh Kumar, and Douglas L. Jones. Scalable stochastic processors. In DATE, 2010. [7] Vicky Wong and Mark Horowitz. Soft error resilience of probabilistic inference applications. In SELSE, 2006. [8]
9. Storage AND NOR NAND Logic Algorithms Е EnerJ Flikker ASPLOS 2011 Kinds of imprecision Relax ISCA 2010 Green PLDI 2010
10. Generality A range of approximation strategies supported with a single abstraction.
11. Critical Non-Critical error-sensitive error-resilient references jump targets JPEG header pixel data neuron weights audio samples video frames
14. Critical Non-Critical error-sensitive error-resilient ✓ ✗ references jump targets JPEG header pixel data neuron weights audio samples video frames
15. Safety Separate critical and non-critical program components. Generality A range of approximation strategies supported with a single abstraction.
16. Java language extension using type annotations Proposed approximate hardware Potential energy savings in existing Java programs
17. Java language extension using type annotations Proposed approximate hardware Potential energy savings in existing Java programs
18. Java language extension using type annotations Type qualifiers: @Approx @Precise & Endorsement Operator overloading Prevention of implicit flows Objects: qualifier polymorphism
19. ✓ ✗ int a = ...; int p = ...; @Approx @Precise p = a; a = p; Type qualifiers
20. endorse( ) ✓ ✗ int a = expensiveCalc(); int p; @Approx @Precise p = a; Endorsement: escape hatch quickChecksum(p); output(p);
21. int a = ...; int p = ...; @Approx @Precise p + p; p + a; a + a; + : @Precise int, @Precise int → @Precise int + : @Approx int, @Approx int → @Approx int Logic approximation: overloading
22. ✗ int a = ...; int p = ...; @Approx @Precise if ( p = 2; } a == 10) { Control flow
23. ✓ int a = ...; int p = ...; @Approx @Precise if ( p = 2; } a == 10 ) { endorse( ) Control flow
24. float mean() { calculate mean } float[] nums = ...; class FloatSet { new @Approx FloatSet() new @Precise FloatSet() } Objects
25. float mean() { calculate mean } float[] nums = ...; class FloatSet { @Context } Objects
26. float mean() { calculate mean } float[] nums = ...; class FloatSet { @Context } @Approx float mean_APPROX() { take mean of first ½ } @Approx FloatSet someSet = ...; someSet.mean();
27. Java language extension using type annotations Proposed approximate hardware Potential energy savings in existing Java programs
28. Storage AND NOR NAND Logic Algorithms Е EnerJ Hypothetical hardware
29. CPU Hypothetical hardware Memory Registers Functional Units Data Cache Lower SRAM supply voltage Lower DRAM refresh rate Reduced VDD Smaller FP mantissa
30. Java language extension using type annotations Proposed approximate hardware Potential energy savings in existing Java programs
31. 0% 25% 50% 75% 100% FFT SOR MonteCarlo SMM LU ZXing jME ImageJ Raytracer 33% 34% 19% 4% 23% 14% 20% 25% 33% Annotated declarations SciMark2 algorithms other kernels full application Annotations are sparse & straightforward to insert
32. 0% 25% 50% 75% 100% FFT SOR MonteCarlo SMM LU ZXing jME ImageJ Raytracer Base Mild Medium Aggressive Total energy used Saved 10%—50% of total execution energy (in simulation)
33. 0% 25% 50% 75% 100% FFT SOR MonteCarlo SMM LU ZXing jME ImageJ Raytracer Base Mild Medium Aggressive Quality-of-service tradeoff: output error “Mild” configuration is a good fit for all Some applications can tolerate more approximation
34. Also in the paper: Formal semantics Noninterference claim Object layout Hardware model Quality-of-service metrics
35. EnerJ: Save energy using programmer controls over execution correctness.