Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory

碩士 === 國立中正大學 === 資訊工程所 === 98 === Recently, demand paging technology with NAND flash memory has been applied widely to embedded system with NAND flash memory. Traditionally, the page fault is a major performance bottleneck in demand paging systems. The key problem is how to reduce the times of page...

Full description

Bibliographic Details
Main Authors: Cyun-Yuan Chen, 陳群元
Other Authors: Rong-Guey Chang
Format: Others
Language:en_US
Published: 2010
Online Access:http://ndltd.ncl.edu.tw/handle/20664733452456127637
id ndltd-TW-098CCU05392048
record_format oai_dc
spelling ndltd-TW-098CCU053920482015-10-13T18:25:32Z http://ndltd.ncl.edu.tw/handle/20664733452456127637 Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory 在需求分頁快閃記憶體上以編譯器輔助程式碼重排之技術 Cyun-Yuan Chen 陳群元 碩士 國立中正大學 資訊工程所 98 Recently, demand paging technology with NAND flash memory has been applied widely to embedded system with NAND flash memory. Traditionally, the page fault is a major performance bottleneck in demand paging systems. The key problem is how to reduce the times of page fault to improve system performance. In order to solve the problem, we proposed the method combine compiler and OS technology to work together. First, it analyzes the ELF(Executable and Linking Format) of the program with compiler, and reconstructs it by code arrangement. The code arrangement can cluster the most execution frequency function code together. The code arrangement will produce the binary mapping information. After code arrangement, we can use the code prefetch strategy to reduce the overall system fault page overhead by it to reduce the I/O latency. The experiment results that the requirement of pages are decreased 3.3% on average. Rong-Guey Chang 張榮貴 2010 學位論文 ; thesis 33 en_US
collection NDLTD
language en_US
format Others
sources NDLTD
description 碩士 === 國立中正大學 === 資訊工程所 === 98 === Recently, demand paging technology with NAND flash memory has been applied widely to embedded system with NAND flash memory. Traditionally, the page fault is a major performance bottleneck in demand paging systems. The key problem is how to reduce the times of page fault to improve system performance. In order to solve the problem, we proposed the method combine compiler and OS technology to work together. First, it analyzes the ELF(Executable and Linking Format) of the program with compiler, and reconstructs it by code arrangement. The code arrangement can cluster the most execution frequency function code together. The code arrangement will produce the binary mapping information. After code arrangement, we can use the code prefetch strategy to reduce the overall system fault page overhead by it to reduce the I/O latency. The experiment results that the requirement of pages are decreased 3.3% on average.
author2 Rong-Guey Chang
author_facet Rong-Guey Chang
Cyun-Yuan Chen
陳群元
author Cyun-Yuan Chen
陳群元
spellingShingle Cyun-Yuan Chen
陳群元
Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
author_sort Cyun-Yuan Chen
title Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
title_short Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
title_full Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
title_fullStr Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
title_full_unstemmed Compiler-Assisted Code Arrangement on Demand Paging with NAND Flash Memory
title_sort compiler-assisted code arrangement on demand paging with nand flash memory
publishDate 2010
url http://ndltd.ncl.edu.tw/handle/20664733452456127637
work_keys_str_mv AT cyunyuanchen compilerassistedcodearrangementondemandpagingwithnandflashmemory
AT chénqúnyuán compilerassistedcodearrangementondemandpagingwithnandflashmemory
AT cyunyuanchen zàixūqiúfēnyèkuàishǎnjìyìtǐshàngyǐbiānyìqìfǔzhùchéngshìmǎzhòngpáizhījìshù
AT chénqúnyuán zàixūqiúfēnyèkuàishǎnjìyìtǐshàngyǐbiānyìqìfǔzhùchéngshìmǎzhòngpáizhījìshù
_version_ 1718032522800005120