Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression
碩士 === 國立中正大學 === 資訊工程研究所 === 88 === Since text compressions are well-known techniques, we propose the construction of code compression schemes for embedded systems based on these techniques. And the issues we concerned are the program size and the energy consumption. In the aspect of pro...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2000
|
Online Access: | http://ndltd.ncl.edu.tw/handle/82150375332190857635 |
id |
ndltd-TW-088CCU00392041 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-088CCU003920412015-10-13T11:50:28Z http://ndltd.ncl.edu.tw/handle/82150375332190857635 Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression 應用於嵌入式系統的高效率多層次字典與選擇式程式碼壓縮 Chin Chia Huang 黃俊嘉 碩士 國立中正大學 資訊工程研究所 88 Since text compressions are well-known techniques, we propose the construction of code compression schemes for embedded systems based on these techniques. And the issues we concerned are the program size and the energy consumption. In the aspect of program size, we bring up two methods. One is a dictionary based compression scheme, which is called multi-level dictionaries. And the second one named as assembling common code sequences, which can be either dictionary based compression or not. The key idea of the multi-level dictionary approach is to separate dictionary into multi-level dictionaries. There are total three dictionary categories in multi-level dictionaries. Besides, we divide the instruction words into three fields to further compress the programs. The assembling common code sequences method is designed for the RISC microprocessors while the multi-level dictionary method is used for the X86 microprocessors. It separates the instruction words into several common code sequences, and these common code sequences are grouped according to their size. Each group of common code sequence has distinct dictionary entries. Finally, the programs can be encoded either by dictionary indices or Huffman codes. As our experiment results show that either the multi-level dictionaries or the assembling common code sequences techniques can efficiently compress programs. In the aspect of power consumption, we propose a selective compression scheme, which allows entire cache block to be compressed. Besides, byte alignment is another technique to save energy. To efficiently compress the program with less energy dissipation, we only focus on the condition to compress entire instruction blocks instead of the instruction words. The key idea of this selective approach is addressed on how to select instruction blocks that are suitable for compression. Thus, we will show the decisions of selecting compressible instruction blocks. Our simulation results indicate that the selective compression scheme can efficiently save energy. Tien-Fu Chen 陳添福 2000 學位論文 ; thesis 72 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立中正大學 === 資訊工程研究所 === 88 === Since text compressions are well-known techniques, we propose the construction of code compression schemes for embedded systems based on these techniques. And the issues we concerned are the program size and the energy consumption.
In the aspect of program size, we bring up two methods. One is a dictionary based compression scheme, which is called multi-level dictionaries. And the second one named as assembling common code sequences, which can be either dictionary based compression or not. The key idea of the multi-level dictionary approach is to separate dictionary into multi-level dictionaries. There are total three dictionary categories in multi-level dictionaries. Besides, we divide the instruction words into three fields to further compress the programs.
The assembling common code sequences method is designed for the RISC microprocessors while the multi-level dictionary method is used for the X86 microprocessors. It separates the instruction words into several common code sequences, and these common code sequences are grouped according to their size. Each group of common code sequence has distinct dictionary entries. Finally, the programs can be encoded either by dictionary indices or Huffman codes. As our experiment results show that either the multi-level dictionaries or the assembling common code sequences techniques can efficiently compress programs.
In the aspect of power consumption, we propose a selective compression scheme, which allows entire cache block to be compressed. Besides, byte alignment is another technique to save energy. To efficiently compress the program with less energy dissipation, we only focus on the condition to compress entire instruction blocks instead of the instruction words. The key idea of this selective approach is addressed on how to select instruction blocks that are suitable for compression. Thus, we will show the decisions of selecting compressible instruction blocks. Our simulation results indicate that the selective compression scheme can efficiently save energy.
|
author2 |
Tien-Fu Chen |
author_facet |
Tien-Fu Chen Chin Chia Huang 黃俊嘉 |
author |
Chin Chia Huang 黃俊嘉 |
spellingShingle |
Chin Chia Huang 黃俊嘉 Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
author_sort |
Chin Chia Huang |
title |
Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
title_short |
Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
title_full |
Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
title_fullStr |
Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
title_full_unstemmed |
Efficient Code Compression for Embedded Systems by Multi-level Dictionaries and Selective Compression |
title_sort |
efficient code compression for embedded systems by multi-level dictionaries and selective compression |
publishDate |
2000 |
url |
http://ndltd.ncl.edu.tw/handle/82150375332190857635 |
work_keys_str_mv |
AT chinchiahuang efficientcodecompressionforembeddedsystemsbymultileveldictionariesandselectivecompression AT huángjùnjiā efficientcodecompressionforembeddedsystemsbymultileveldictionariesandselectivecompression AT chinchiahuang yīngyòngyúqiànrùshìxìtǒngdegāoxiàolǜduōcéngcìzìdiǎnyǔxuǎnzéshìchéngshìmǎyāsuō AT huángjùnjiā yīngyòngyúqiànrùshìxìtǒngdegāoxiàolǜduōcéngcìzìdiǎnyǔxuǎnzéshìchéngshìmǎyāsuō |
_version_ |
1716849101398605824 |