On the High-level Synthesis of Data Path
博士 === 國立成功大學 === 電機工程研究所 === 82 === In this dissertation, we discuss the problems and a system for high-level pipelined data path synthesis. The main tasks for high-level synthesis are scheduling and allocation. Given a data flow graph, a scheduling pr...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
1993
|
Online Access: | http://ndltd.ncl.edu.tw/handle/61072319802618333106 |
id |
ndltd-TW-082NCKU0442125 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-082NCKU04421252015-10-13T15:36:51Z http://ndltd.ncl.edu.tw/handle/61072319802618333106 On the High-level Synthesis of Data Path 高階合成資料路徑之研究 Yuan-Long Jeang 蔣元隆 博士 國立成功大學 電機工程研究所 82 In this dissertation, we discuss the problems and a system for high-level pipelined data path synthesis. The main tasks for high-level synthesis are scheduling and allocation. Given a data flow graph, a scheduling program is used for scheduling each operation to an appropriate step such that the hardware units can be shared as much as possible. Then, in the allocation phase, the registers, buses, and function units are allocated to specific values , transfers, and operations respectively such that the numbers of connections and multiplexers are minimum. In Chapter 1, the motivation and terminologies are introduced . In Chapter 2, we discuss current scheduling techniques for pipelined data path synthesis. In Chapter 3, we discuss the essential properties of allocation problems and a tabu-search based program is presented and developed for either pipelined and non-pipelined data path synthesis . In Chapter 4, a synthesizer for nested loop is presented based on a modified Pipelined List Scheduler and the tabu- search based allocation program. In Chapter 5, conclusions are made and future work is identified. Jhing-Fa Wang, Jau-Yien Lee 王駿發, 李肇嚴 1993 學位論文 ; thesis 94 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
博士 === 國立成功大學 === 電機工程研究所 === 82 === In this dissertation, we discuss the problems and a system for
high-level pipelined data path synthesis. The main tasks for
high-level synthesis are scheduling and allocation. Given a
data flow graph, a scheduling program is used for scheduling
each operation to an appropriate step such that the hardware
units can be shared as much as possible. Then, in the
allocation phase, the registers, buses, and function units are
allocated to specific values , transfers, and operations
respectively such that the numbers of connections and
multiplexers are minimum. In Chapter 1, the motivation and
terminologies are introduced . In Chapter 2, we discuss current
scheduling techniques for pipelined data path synthesis. In
Chapter 3, we discuss the essential properties of allocation
problems and a tabu-search based program is presented and
developed for either pipelined and non-pipelined data path
synthesis . In Chapter 4, a synthesizer for nested loop is
presented based on a modified Pipelined List Scheduler and the
tabu- search based allocation program. In Chapter 5,
conclusions are made and future work is identified.
|
author2 |
Jhing-Fa Wang, Jau-Yien Lee |
author_facet |
Jhing-Fa Wang, Jau-Yien Lee Yuan-Long Jeang 蔣元隆 |
author |
Yuan-Long Jeang 蔣元隆 |
spellingShingle |
Yuan-Long Jeang 蔣元隆 On the High-level Synthesis of Data Path |
author_sort |
Yuan-Long Jeang |
title |
On the High-level Synthesis of Data Path |
title_short |
On the High-level Synthesis of Data Path |
title_full |
On the High-level Synthesis of Data Path |
title_fullStr |
On the High-level Synthesis of Data Path |
title_full_unstemmed |
On the High-level Synthesis of Data Path |
title_sort |
on the high-level synthesis of data path |
publishDate |
1993 |
url |
http://ndltd.ncl.edu.tw/handle/61072319802618333106 |
work_keys_str_mv |
AT yuanlongjeang onthehighlevelsynthesisofdatapath AT jiǎngyuánlóng onthehighlevelsynthesisofdatapath AT yuanlongjeang gāojiēhéchéngzīliàolùjìngzhīyánjiū AT jiǎngyuánlóng gāojiēhéchéngzīliàolùjìngzhīyánjiū |
_version_ |
1717767144212529152 |