國立虎尾科技大學 |

High-level synthesis of polynomial datapaths using finite integer algebras.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	High-level synthesis of polynomial datapaths using finite integer algebras./
Author:	Gopalakrishnan, Sivaram.
Description:	1 online resource (134 pages)
Notes:	Source: Dissertation Abstracts International, Volume: 69-11, Section: B, page: 6914.
Contained By:	Dissertation Abstracts International69-11B.
Subject:	Computer science. -
Online resource:	click for full text (PQDT)
ISBN:	9780549921202

High-level synthesis of polynomial datapaths using finite integer algebras.
Gopalakrishnan, Sivaram.

High-level synthesis of polynomial datapaths using finite integer algebras. - 1 online resource (134 pages)

Source: Dissertation Abstracts International, Volume: 69-11, Section: B, page: 6914.

Thesis (Ph.D.)--The University of Utah, 2008.

Includes bibliographical references

High-level descriptions of integer datapaths that implement polynomial arithmetic are found in many practical applications, such as in Digital Signal Processing (DSP) for audio, video and multimedia applications. The growing demand for such applications has resulted in the need for efficient CAD support for their synthesis at high-level. Conventional high-level synthesis tools are quite adept at capturing hardware description language (HDL) models and mapping them into control/data-flow graphs (CDFGs), and performing optimizations such as: scheduling, resource allocation and sharing, binding, and retiming. For DSP designs that generally implement data-flow intensive operations in fixed-point arithmetic, where algebraic computations are performed with finite precision, these tools lack the mathematical wherewithal to perform sophisticated algebraic manipulation. This dissertation addresses the synthesis and optimization of arithmetic designs with specific regards to finite precision.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780549921202Subjects--Topical Terms:

573171
Computer science.
Index Terms--Genre/Form:

554714
Electronic books.

High-level synthesis of polynomial datapaths using finite integer algebras.
LDR:03665ntm a2200325Ki 4500 001 915982
005 20180907134545.5
006 m o u
007 cr mn||||a|a||
008 190606s2008 xx obm 000 0 eng d
020 $a 9780549921202
035 $a (MiAaPQ)AAI3338136
035 $a AAI3338136
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Gopalakrishnan, Sivaram. $3 1189544
245 1 0 $a High-level synthesis of polynomial datapaths using finite integer algebras.
264 0 $c 2008
300 $a 1 online resource (134 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 69-11, Section: B, page: 6914.
502 $a Thesis (Ph.D.)--The University of Utah, 2008.
504 $a Includes bibliographical references
520 $a High-level descriptions of integer datapaths that implement polynomial arithmetic are found in many practical applications, such as in Digital Signal Processing (DSP) for audio, video and multimedia applications. The growing demand for such applications has resulted in the need for efficient CAD support for their synthesis at high-level. Conventional high-level synthesis tools are quite adept at capturing hardware description language (HDL) models and mapping them into control/data-flow graphs (CDFGs), and performing optimizations such as: scheduling, resource allocation and sharing, binding, and retiming. For DSP designs that generally implement data-flow intensive operations in fixed-point arithmetic, where algebraic computations are performed with finite precision, these tools lack the mathematical wherewithal to perform sophisticated algebraic manipulation. This dissertation addresses the synthesis and optimization of arithmetic designs with specific regards to finite precision.
520 $a In fixed-point arithmetic, polynomials are implemented over specific bit-vector sizes. A bit-vector of size m represents integer values from 0 to 2m--1 (or integers reduced modulo 2m). This implies that finite word-length (m) bit-vector arithmetic manifests itself as algebra over finite integer rings of residue classes Z2m . Contemporary algebra-based synthesis techniques model polynomial computations over unique factorization domains (UFDs) such as fields, Euclidean and integral domains. However, the finite ring Z2m , which is a nonunique factorization domain (nonUFD), does not fall in this category, rendering contemporary approaches inapplicable.
520 $a In this dissertation, the nonUFD nature of the ring Z2m is viewed, not as a liability, but as a resource that offers potential for optimization of bit-vector arithmetic. We model the datapath as polynomial functions over rings of the type Z2m , and then exploit the number-theoretic and algebraic properties of such rings, to perform high-level optimization of finite-precision polynomial arithmetic. Systematic algorithmic procedures are developed and integrated within a CAD-based framework to perform the optimization. Experiments conducted over a variety of benchmarks demonstrate significant area savings as a result of our optimization. We analyze the experimental results and review the advantages and limitations of our approaches. Finally, we suggest some future research directions based on the work presented in this research.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Computer science. $3 573171
655 7 $a Electronic books. $2 local $3 554714
690 $a 0984
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The University of Utah. $3 1182029
773 0 $t Dissertation Abstracts International $g 69-11B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3338136 $z click for full text (PQDT)