國立高雄大學圖資館 |

Language: English

Back

Data orchestration in deep learning ...

Krishna, Tushar,

Data orchestration in deep learning accelerators

Record Type:	Electronic resources : Monograph/item
Title/Author:	Data orchestration in deep learning acceleratorsTushar Krishna, Hyoukjun Kwon, Angshuman Parashar, Michael Pellauer, Ananda Samajdar.
Author:	Krishna, Tushar,
other author:	Kwon, Hyoukjun,
Description:	1 online resource (166 p.)
Subject:	Neural networks (Computer science)
Online resource:	https://portal.igpublish.com/iglibrary/search/MCPB0006576.html
ISBN:	9781681738697

Data orchestration in deep learning accelerators
Krishna, Tushar,

Data orchestration in deep learning accelerators[electronic resource] /Tushar Krishna, Hyoukjun Kwon, Angshuman Parashar, Michael Pellauer, Ananda Samajdar. - 1 online resource (166 p.) - Synthesis lectures on computer architecture ;52. - Synthesis lectures on computer architecture ;52..

Includes bibliographical references (pages 131-143).

Access restricted to authorized users and institutions.

This Synthesis Lecture focuses on techniques for efficient data orchestration within DNN accelerators. The End of Moore's Law, coupled with the increasing growth in deep learning and other AI applications has led to the emergence of custom Deep Neural Network (DNN) accelerators for energy-efficient inference on edge devices. Modern DNNs have millions of hyper parameters and involve billions of computations this necessitates extensive data movement from memory to on-chip processing engines. It is well known that the cost of data movement today surpasses the cost of the actual computation therefore, DNN accelerators require careful orchestration of data across on-chip compute, network, and memory elements to minimize the number of accesses to external DRAM. The book covers DNN dataflows, data reuse, buffer hierarchies, networks-on-chip, and automated design-space exploration. It concludes with data orchestration challenges with compressed and sparse DNNs and future trends. The target audience is students, engineers, and researchers interested in designing high-performance and low-energy accelerators for DNN inference

Mode of access: World Wide Web.

ISBN: 9781681738697Subjects--Topical Terms:

181982
Neural networks (Computer science)
Index Terms--Genre/Form:

214472
Electronic books.

LC Class. No.: Q342

Dewey Class. No.: 006.3

Data orchestration in deep learning accelerators
LDR:02284nmm a2200301 i 4500 001 603281
006 m eo d
007 cr cn |||m|||a
008 211117t20202020cau ob 000 0 eng d
020 $a 9781681738697
020 $a 9781681738703
020 $a 9781681738710
035 $a MCPB0006576
040 $a iG Publishing $b eng $c iG Publishing $e rda
050 0 0 $a Q342
082 0 0 $a 006.3
100 1 $a Krishna, Tushar, $e author. $3 899511
245 1 0 $a Data orchestration in deep learning accelerators $h [electronic resource] / $c Tushar Krishna, Hyoukjun Kwon, Angshuman Parashar, Michael Pellauer, Ananda Samajdar.
264 1 $a San Rafael, California : $b Morgan & Claypool Publishers, $c 2020.
264 4 $c ©2020
300 $a 1 online resource (166 p.)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
490 1 $a Synthesis lectures on computer architecture ; $v 52
504 $a Includes bibliographical references (pages 131-143).
506 $a Access restricted to authorized users and institutions.
520 3 $a This Synthesis Lecture focuses on techniques for efficient data orchestration within DNN accelerators. The End of Moore's Law, coupled with the increasing growth in deep learning and other AI applications has led to the emergence of custom Deep Neural Network (DNN) accelerators for energy-efficient inference on edge devices. Modern DNNs have millions of hyper parameters and involve billions of computations this necessitates extensive data movement from memory to on-chip processing engines. It is well known that the cost of data movement today surpasses the cost of the actual computation therefore, DNN accelerators require careful orchestration of data across on-chip compute, network, and memory elements to minimize the number of accesses to external DRAM. The book covers DNN dataflows, data reuse, buffer hierarchies, networks-on-chip, and automated design-space exploration. It concludes with data orchestration challenges with compressed and sparse DNNs and future trends. The target audience is students, engineers, and researchers interested in designing high-performance and low-energy accelerators for DNN inference
538 $a Mode of access: World Wide Web.
650 0 $a Neural networks (Computer science) $3 181982
650 0 $a Machine learning. $3 188639
650 0 $a Data flow computing. $3 348292
655 4 $a Electronic books. $2 local. $3 214472
700 1 $a Kwon, Hyoukjun, $e author. $3 899512
700 1 $a Parashar, Angshuman, $e author. $3 899513
700 1 $a Pellauer, Michael, $e author. $3 899514
700 1 $a Samajdar, Ananda, $e author. $3 899515
830 0 $a Synthesis lectures on computer architecture ; $v 52. $3 899516
856 4 0 $u https://portal.igpublish.com/iglibrary/search/MCPB0006576.html