Reply to thread

Message

<blockquote data-quote="MJGraf" data-source="post: 1004959" data-attributes="member: 17886">...and just for reference:&nbsp;I also tried to add SQLite's shared cache mode (<a href="http://www.sqlite.org/sharedcache.html" target="_blank">http://www.sqlite.org/sharedcache.html</a>) and SQLite's memory mapped i/o (<a href="http://www.sqlite.org/mmap.html" target="_blank">http://www.sqlite.org/mmap.html</a>) via the respective PRAGMA Statements in every SQLiteConnection to see, whether these have any positive impact on the speed of the SQLiteDatabase plugin (in particular with respect to share imports).&nbsp;Both settings did not have any effect on speed at least from the tests I ran (i.e. importing large shares), so I removed the respective statements again.&nbsp;Fot memory mapped i/o this may result from the fact that if you set e.g. a memory size to use for mapping of 512MB (which is then actually used in RAM for this), only the first 512MB of the database are mapped. For every database access beyond 512MB in your database, the file is accessed regularly, i.e. without memory mapping. My database is 1.7GB in size, so I also tried to use 2GB of my RAM for memory mapping (which is the maximum value for SQLite), but again no speed improvements (apart from the fact that using 2GB of RAM just for speed improvements in the database is a real overkill...)&nbsp;Just wanted to mention this here for documentation purposes. At least from my perspective no need to investigate this any further...&nbsp;Michael</blockquote>

[QUOTE="MJGraf, post: 1004959, member: 17886"] ...and just for reference: I also tried to add SQLite's shared cache mode ([url]http://www.sqlite.org/sharedcache.html[/url]) and SQLite's memory mapped i/o ([url]http://www.sqlite.org/mmap.html[/url]) via the respective PRAGMA Statements in every SQLiteConnection to see, whether these have any positive impact on the speed of the SQLiteDatabase plugin (in particular with respect to share imports). Both settings did not have any effect on speed at least from the tests I ran (i.e. importing large shares), so I removed the respective statements again. Fot memory mapped i/o this may result from the fact that if you set e.g. a memory size to use for mapping of 512MB (which is then actually used in RAM for this), only the first 512MB of the database are mapped. For every database access beyond 512MB in your database, the file is accessed regularly, i.e. without memory mapping. My database is 1.7GB in size, so I also tried to use 2GB of my RAM for memory mapping (which is the maximum value for SQLite), but again no speed improvements (apart from the fact that using 2GB of RAM just for speed improvements in the database is a real overkill...) Just wanted to mention this here for documentation purposes. At least from my perspective no need to investigate this any further... Michael [/QUOTE]

Verification