Washington University

Similar documents
Washington University School of Engineering and Applied Science. Power Consumption of Customized Numerical Representations for Audio Signal Processing

Multi-Modal Communication

High-Assurance Security/Safety on HPEC Systems: an Oxymoron?

Running CyberCIEGE on Linux without Windows

4. Lessons Learned in Introducing MBSE: 2009 to 2012

Service Level Agreements: An Approach to Software Lifecycle Management. CDR Leonard Gaines Naval Supply Systems Command 29 January 2003

An Efficient Architecture for Ultra Long FFTs in FPGAs and ASICs

DoD Common Access Card Information Brief. Smart Card Project Managers Group

ARINC653 AADL Annex Update

Empirically Based Analysis: The DDoS Case

FUDSChem. Brian Jordan With the assistance of Deb Walker. Formerly Used Defense Site Chemistry Database. USACE-Albuquerque District.

Setting the Standard for Real-Time Digital Signal Processing Pentek Seminar Series. Digital IF Standardization

MODELING AND SIMULATION OF LIQUID MOLDING PROCESSES. Pavel Simacek Center for Composite Materials University of Delaware

Using Model-Theoretic Invariants for Semantic Integration. Michael Gruninger NIST / Institute for Systems Research University of Maryland

Dana Sinno MIT Lincoln Laboratory 244 Wood Street Lexington, MA phone:

Distributed Real-Time Embedded Video Processing

Topology Control from Bottom to Top

Information, Decision, & Complex Networks AFOSR/RTC Overview

Kathleen Fisher Program Manager, Information Innovation Office

Towards a Formal Pedigree Ontology for Level-One Sensor Fusion

An Update on CORBA Performance for HPEC Algorithms. Bill Beckwith Objective Interface Systems, Inc.

COMPUTATIONAL FLUID DYNAMICS (CFD) ANALYSIS AND DEVELOPMENT OF HALON- REPLACEMENT FIRE EXTINGUISHING SYSTEMS (PHASE II)

73rd MORSS CD Cover Page UNCLASSIFIED DISCLOSURE FORM CD Presentation

Requirements for Scalable Application Specific Processing in Commercial HPEC

2013 US State of Cybercrime Survey

Web Site update. 21st HCAT Program Review Toronto, September 26, Keith Legg

73rd MORSS CD Cover Page UNCLASSIFIED DISCLOSURE FORM CD Presentation

Technological Advances In Emergency Management

Vision Protection Army Technology Objective (ATO) Overview for GVSET VIP Day. Sensors from Laser Weapons Date: 17 Jul 09 UNCLASSIFIED

High-Performance Linear Algebra Processor using FPGA

ENVIRONMENTAL MANAGEMENT SYSTEM WEB SITE (EMSWeb)

U.S. Army Research, Development and Engineering Command (IDAS) Briefer: Jason Morse ARMED Team Leader Ground System Survivability, TARDEC

DoD M&S Project: Standardized Documentation for Verification, Validation, and Accreditation

SINOVIA An open approach for heterogeneous ISR systems inter-operability

CASE STUDY: Using Field Programmable Gate Arrays in a Beowulf Cluster

A Distributed Parallel Processing System for Command and Control Imagery

75th Air Base Wing. Effective Data Stewarding Measures in Support of EESOH-MIS

Wireless Connectivity of Swarms in Presence of Obstacles

ASPECTS OF USE OF CFD FOR UAV CONFIGURATION DESIGN

Space and Missile Systems Center

Data Reorganization Interface

A Review of the 2007 Air Force Inaugural Sustainability Report

Space and Missile Systems Center

Concept of Operations Discussion Summary

Use of the Polarized Radiance Distribution Camera System in the RADYO Program

ASSESSMENT OF A BAYESIAN MODEL AND TEST VALIDATION METHOD

CENTER FOR ADVANCED ENERGY SYSTEM Rutgers University. Field Management for Industrial Assessment Centers Appointed By USDOE

Balancing Transport and Physical Layers in Wireless Ad Hoc Networks: Jointly Optimal Congestion Control and Power Control

The C2 Workstation and Data Replication over Disadvantaged Tactical Communication Links

INTEGRATING LOCAL AND GLOBAL NAVIGATION IN UNMANNED GROUND VEHICLES

Architecting for Resiliency Army s Common Operating Environment (COE) SERC

Sparse Linear Solver for Power System Analyis using FPGA

Accuracy of Computed Water Surface Profiles

Lessons Learned in Adapting a Software System to a Micro Computer

LARGE AREA, REAL TIME INSPECTION OF ROCKET MOTORS USING A NOVEL HANDHELD ULTRASOUND CAMERA

Shallow Ocean Bottom BRDF Prediction, Modeling, and Inversion via Simulation with Surface/Volume Data Derived from X-Ray Tomography

Monte Carlo Techniques for Estimating Power in Aircraft T&E Tests. Todd Remund Dr. William Kitto EDWARDS AFB, CA. July 2011

ATCCIS Replication Mechanism (ARM)

Ross Lazarus MB,BS MPH

Advanced Numerical Methods for Numerical Weather Prediction

SURVIVABILITY ENHANCED RUN-FLAT

David W. Hyde US Army Engineer Waterways Experiment Station Vicksburg, Mississippi ABSTRACT

The extreme Adaptive DSP Solution to Sensor Data Processing

Data Warehousing. HCAT Program Review Park City, UT July Keith Legg,

NEW FINITE ELEMENT / MULTIBODY SYSTEM ALGORITHM FOR MODELING FLEXIBLE TRACKED VEHICLES

Edwards Air Force Base Accelerates Flight Test Data Analysis Using MATLAB and Math Works. John Bourgeois EDWARDS AFB, CA. PRESENTED ON: 10 June 2010

Android: Call C Functions with the Native Development Kit (NDK)

Using Templates to Support Crisis Action Mission Planning

Computer Aided Munitions Storage Planning

Data Replication in the CNR Environment:

DEVELOPMENT OF A NOVEL MICROWAVE RADAR SYSTEM USING ANGULAR CORRELATION FOR THE DETECTION OF BURIED OBJECTS IN SANDY SOILS

Col Jaap de Die Chairman Steering Committee CEPA-11. NMSG, October

Parallel Matlab: RTExpress on 64-bit SGI Altix with SCSL and MPT

Development and Test of a Millimetre-Wave ISAR Images ATR System

Parallelization of a Electromagnetic Analysis Tool

C2-Simulation Interoperability in NATO

WAITING ON MORE THAN 64 HANDLES

Dr. Stuart Dickinson Dr. Donald H. Steinbrecher Naval Undersea Warfare Center, Newport, RI May 10, 2011

Corrosion Prevention and Control Database. Bob Barbin 07 February 2011 ASETSDefense 2011

BUPT at TREC 2009: Entity Track

Energy Security: A Global Challenge

Basing a Modeling Environment on a General Purpose Theorem Prover

75 TH MORSS CD Cover Page. If you would like your presentation included in the 75 th MORSS Final Report CD it must :

Report Documentation Page

Virtual Prototyping with SPEED

SETTING UP AN NTP SERVER AT THE ROYAL OBSERVATORY OF BELGIUM

Exploring the Query Expansion Methods for Concept Based Representation

2011 NNI Environment, Health, and Safety Research Strategy

New Dimensions in Microarchitecture Harnessing 3D Integration Technologies

QuanTM Architecture for Web Services

Speaker Verification Using SVM

Nationwide Automatic Identification System (NAIS) Overview. CG 939 Mr. E. G. Lockhart TEXAS II Conference 3 Sep 2008

Enhanced Predictability Through Lagrangian Observations and Analysis

A Methodology for End-to-End Evaluation of Arabic Document Image Processing Software

Agile Modeling of Component Connections for Simulation and Design of Complex Vehicle Structures

Dr. ing. Rune Aasgaard SINTEF Applied Mathematics PO Box 124 Blindern N-0314 Oslo NORWAY

Uniform Tests of File Converters Using Unit Cubes

Advanced Numerical Methods for Numerical Weather Prediction

Using the SORASCS Prototype Web Portal

M&S Strategic Initiatives to Support Test & Evaluation

Transcription:

Washington University School of Engineering and Applied Science Power Consumption of Customized Numerical Representations for Audio Signal Processing Roger Chamberlain, Yen Hsiang Chew, Varuna DeAlwis, Eric Hemmeter,John Lockwood, Robert Morley, Ed Richter, Jason White, and Huakai Zhang This research is supported by the NIH under grant 1R4-3DC04028-02. HPEC 2002 1

Report Documentation Page Form Approved OMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. 1. REPORT DATE 24 SEP 2002 2. REPORT TYPE N/A 3. DATES COVERED - 4. TITLE AND SUBTITLE Power Consumption of Customized Numerical Representations for Audio Signal Processing 5a. CONTRACT NUMBER 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Washington University School of Engineering and Applied Science 8. PERFORMING ORGANIZATION REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR S ACRONYM(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release, distribution unlimited 13. SUPPLEMENTARY NOTES Also see ADM001473, The original document contains color images. 14. ABSTRACT 15. SUBJECT TERMS 11. SPONSOR/MONITOR S REPORT NUMBER(S) 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT UU a. REPORT unclassified b. ABSTRACT unclassified c. THIS PAGE unclassified 18. NUMBER OF PAGES 21 19a. NAME OF RESPONSIBLE PERSON Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18

Outline Audio Signal Requirements Customized Numerical Representations SNR and Dynamic Range Design of Computation Structures Power Consumption Results Summary and Conclusions HPEC 2002 2

Audio Signal Applications Music MP3 players Speech communications equipment hearing aids (our target application) Signal requirements to understand speech ~30 db SNR over entire dynamic range ~100 db dynamic range Power consumption critical for all of above HPEC 2002 3

Customized Numerical Representations 16-bit integer is traditional for audio 90 db dynamic range, SNR from 0 to 90 db Logarithmic representation more closely mimics human perception Loudness response is highly non-linear SNR is relatively constant across dynamic range Floating point representations are partially logarithmic and partially linear 32-bit IEEE standard is more than is needed Tailor choice for number of bits in exponent and mantissa to needs of application HPEC 2002 4

SNR and Dynamic Range SNR (db) = 20 log 10 2 V i V + 1 i 2 V 12 i Dynamic Range (db) = 20 log 10 V V max min HPEC 2002 5

16-bit Integer SNR SNR (db) 100 90 80 70 60 50 40 30 20 10 0 0.00001 0.0001 0.001 0.01 0.1 1 Input Value (Vi) HPEC 2002 6

9-bit Logarithmic SNR 60 50 40 SNR (db) 30 20 10 0 1E-07 0.000001 0.00001 0.0001 0.001 0.01 0.1 1 Input Value (Vi) HPEC 2002 7

Floating Point SNR 5-bit exponent 4, 5, and 6-bit mantissa SNR (db) 60 50 40 30 20 10 0 1E-10 0.00000001 0.000001 0.0001 0.01 1 Input Value (Vi) FP 5-6 FP 5-5 FP 5-4 HPEC 2002 8

400 6-bit exp. Dynamic Range 350 300 Range (db) 250 200 150 100 5-bit exp. log. 4-bit exp. int. 50 0 6-6 FP 6-5 FP 6-4 FP 5-6 FP 5-5 FP 5-4 FP 9-bit Log 4-6 FP 4-5 FP 4-4 FP 16-bit Integer Representation HPEC 2002 9

Hearing Aid Architecture Most of the computations the hearing aid performs are multiply-accumulates. HPEC 2002 10

Floating Point MAC Design Traditional structure for floating point hardware computations Perl script generates synthesizable VHDL code for specific exponent and mantissa size Small size of mantissa implies lower power multiplier hardware HPEC 2002 11

Floating Point MAC Xi_j/Cj(4 downto 0) Yi(4 downto 0) Xi_j Mant * Mant Normalize Shifter Acc Mant S Mant Normalize Register Yi Cj Xi_j/Cj(9 downto 5) Exp + Exp Normalize Exp Sub Yi(10) Exp Normalize Yi(9 downto 5) S Xi_j/Cj(10) S Clk Latch Enable HPEC 2002 12 Reset

Logarithmic MAC Design Multiply function provided by an adder: log( A B) = log( A) + log( B) Addition function exploits the following relationship: log ( A + B) = log( A) + log 1+ Last term implemented via a lookup table B A HPEC 2002 13

9-bit Logarithmic MAC Xi_j Cj Xi_j/Cj(8 downto 0) * Multiplied Result Look-up Table Add Result or Pass-through Register Yi Xi_j/Cj (9) s s Comparator Clk Latch Reset Enable HPEC 2002 14

Verification Via FPX Platform Audio Data FPX Fieldprogrammable Port Extender Audio Data HPEC 2002 15

FPX Platform HPEC 2002 16

Layout AMI 0.5 ìm process ADK library from Mentor Graphics HEP 5-5 floating point MAC is shown HPEC 2002 17

Power Estimation via Simulation Simulate using Mentor Graphics MACH-PA Spice-level simulation tool Driven by extracted layout Focus on Multiply-Accumulate units Random input vectors Simulation provides current usage P = IV provides power results HPEC 2002 18

Power Consumption Average Power (mw) 80 60 40 20 0 70.5 16-bit Integer 10.9 9.6 10.15 8.5 2.7 1-5-5 1-5-4 1-4-5 1-4-4 9-bit Log Numerical Representation HPEC 2002 19

Summary and Conclusions Customizing a numerical representation to the specific needs of an application can have tangible benefits Several 9 or 10-bit representations have improved SNR and dynamic range for audio speech applications relative to traditional 16- bit integers Both customized floating point and logarithmic representations have been considered Power savings are significant HPEC 2002 20

For Further Information http://www.ccrc.wustl.edu/hearingaid or hearingaid@ccrc.wustl.edu HPEC 2002 21