Area and delay estimation for constraint-driven high-level synthesis

Bibliographic Details
Main Author: Nourani-Dargiri, Mehrdad
Language:English
Published: Case Western Reserve University School of Graduate Studies / OhioLINK 1994
Subjects:
Online Access:http://rave.ohiolink.edu/etdc/view?acc_num=case1057603424
id ndltd-OhioLink-oai-etd.ohiolink.edu-case1057603424
record_format oai_dc
spelling ndltd-OhioLink-oai-etd.ohiolink.edu-case10576034242021-08-03T05:31:04Z Area and delay estimation for constraint-driven high-level synthesis Nourani-Dargiri, Mehrdad constraint driven synthesis The flurry of activity in high level synthesis and its increasing popularity as a research topic, both in industry and academia, is a natural consequence of the progress in VLSI design technology and methodology. High level synthesis (HLS) fills the gap between behavioral level and layout level by automatically generating an RTL datapath realization from a behavioral description. The actual circuit layout can be generated later from the RTL datapath using a silicon compiler. A common characteristic of most HLS systems is the lack of consideration to the logic and layout synthesis aspects during the HLS process. The designs generated by these systems may or may not satisfy the initial design requirements and thus the design may need to be changed or modified. One method to address this deficiency is to use estimators, as a bridge between high and low levels, during HLS. The estimators can be iteratively invoked at high levels (e.g. scheduling and allocation) to provide a guiding mechanism to search the design space, which in turn leads to improved design quality and shorter design turnaround time. Motivated by the above observations, in this dissertation we present methods to cover the gap between high and low levels of design hierarchy, including: (1) Scheduling and mixed scheduling-allocation algorithms under time and resource constraints based on Liapunov stability theorem; (2) A neural network based scheduling as an alternative for implementing on parallel machines; (3) A layout estimation method which uses non-probabilistic analytical formulas in a constructive algorithm to estimate the overall area of a datapath; (4) A delay estimation algorithm which identifies false paths at the RTL and computes the static and dynamic critical delay in a datapath. Incorporation of these tools into an experimental HLS system, SYNTEST, has been completed with the overall objective of generating a structural VHDL description of a self-testable RTL datapath from a given behavioral VHDL specification. 1994 English text Case Western Reserve University School of Graduate Studies / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=case1057603424 http://rave.ohiolink.edu/etdc/view?acc_num=case1057603424 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.
collection NDLTD
language English
sources NDLTD
topic constraint driven synthesis
spellingShingle constraint driven synthesis
Nourani-Dargiri, Mehrdad
Area and delay estimation for constraint-driven high-level synthesis
author Nourani-Dargiri, Mehrdad
author_facet Nourani-Dargiri, Mehrdad
author_sort Nourani-Dargiri, Mehrdad
title Area and delay estimation for constraint-driven high-level synthesis
title_short Area and delay estimation for constraint-driven high-level synthesis
title_full Area and delay estimation for constraint-driven high-level synthesis
title_fullStr Area and delay estimation for constraint-driven high-level synthesis
title_full_unstemmed Area and delay estimation for constraint-driven high-level synthesis
title_sort area and delay estimation for constraint-driven high-level synthesis
publisher Case Western Reserve University School of Graduate Studies / OhioLINK
publishDate 1994
url http://rave.ohiolink.edu/etdc/view?acc_num=case1057603424
work_keys_str_mv AT nouranidargirimehrdad areaanddelayestimationforconstraintdrivenhighlevelsynthesis
_version_ 1719421175650582528