Implementation considerations for "soft" embedded programmable logic cores

As integrated circuits become increasingly more complex and expensive, the ability to make post-fabrication changes will become much more attractive. This ability can be realized using programmable logic cores. Currently, such cores are available from vendors in the form of "hard" macro la...

Full description

Bibliographic Details
Main Author: Wu, James Cheng-Huan
Language:English
Published: 2009
Online Access:http://hdl.handle.net/2429/17079
id ndltd-UBC-oai-circle.library.ubc.ca-2429-17079
record_format oai_dc
spelling ndltd-UBC-oai-circle.library.ubc.ca-2429-170792018-01-05T17:38:44Z Implementation considerations for "soft" embedded programmable logic cores Wu, James Cheng-Huan As integrated circuits become increasingly more complex and expensive, the ability to make post-fabrication changes will become much more attractive. This ability can be realized using programmable logic cores. Currently, such cores are available from vendors in the form of "hard" macro layouts. An alternative approach for fine-grain programmability is possible: vendors supply an RTL version of their programmable logic fabric that can be synthesized using standard cells. Although this technique may suffer in terms of speed, density, and power overhead, the task of integrating such cores is far easier than the task of integrating "hard" cores into an ASIC or SoC. When the required amount of programmable logic is small, this ease of use may be more important than the increased overhead. In this thesis, we identify potential implementation issues associated with such cores, and investigate in depth the area, speed and power overhead of using this approach. Based on this investigation, we attempt to improve the performance of programmable cores created in this manner. Using a test-chip implementation, we identify three main issues: core size selection, I/O connections, and clock-tree synthesis. Compared to a non-programmable design, the soft core approach exhibited an average area overhead of 200X, speed overhead of 10X, and power overhead of 15OX. These numbers are high but expected, given that the approach is subject to limitations of the standard cell library elements of the ASIC flow, which are not optimized for use with programmable logic. Applied Science, Faculty of Electrical and Computer Engineering, Department of Graduate 2009-12-23T00:12:41Z 2009-12-23T00:12:41Z 2004 2004-11 Text Thesis/Dissertation http://hdl.handle.net/2429/17079 eng For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.
collection NDLTD
language English
sources NDLTD
description As integrated circuits become increasingly more complex and expensive, the ability to make post-fabrication changes will become much more attractive. This ability can be realized using programmable logic cores. Currently, such cores are available from vendors in the form of "hard" macro layouts. An alternative approach for fine-grain programmability is possible: vendors supply an RTL version of their programmable logic fabric that can be synthesized using standard cells. Although this technique may suffer in terms of speed, density, and power overhead, the task of integrating such cores is far easier than the task of integrating "hard" cores into an ASIC or SoC. When the required amount of programmable logic is small, this ease of use may be more important than the increased overhead. In this thesis, we identify potential implementation issues associated with such cores, and investigate in depth the area, speed and power overhead of using this approach. Based on this investigation, we attempt to improve the performance of programmable cores created in this manner. Using a test-chip implementation, we identify three main issues: core size selection, I/O connections, and clock-tree synthesis. Compared to a non-programmable design, the soft core approach exhibited an average area overhead of 200X, speed overhead of 10X, and power overhead of 15OX. These numbers are high but expected, given that the approach is subject to limitations of the standard cell library elements of the ASIC flow, which are not optimized for use with programmable logic. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
author Wu, James Cheng-Huan
spellingShingle Wu, James Cheng-Huan
Implementation considerations for "soft" embedded programmable logic cores
author_facet Wu, James Cheng-Huan
author_sort Wu, James Cheng-Huan
title Implementation considerations for "soft" embedded programmable logic cores
title_short Implementation considerations for "soft" embedded programmable logic cores
title_full Implementation considerations for "soft" embedded programmable logic cores
title_fullStr Implementation considerations for "soft" embedded programmable logic cores
title_full_unstemmed Implementation considerations for "soft" embedded programmable logic cores
title_sort implementation considerations for "soft" embedded programmable logic cores
publishDate 2009
url http://hdl.handle.net/2429/17079
work_keys_str_mv AT wujameschenghuan implementationconsiderationsforsoftembeddedprogrammablelogiccores
_version_ 1718590427273101312