1. Is anyone else on the list using Visual Studio 2005?  It would be
handy to see if they got similar results with the test application or
not, that would rule out the build environment to an extent.  
2. And the other thing to try would be if anyone has a fairly meaty test
database they don't mind sharing that I could fling at my test
application to try and rule out the data?

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Dennis Cote
Sent: Thursday, August 21, 2008 4:23 PM
To: General Discussion of SQLite Database
Subject: Re: [sqlite] Reducing SQLite Memory footprint(!)

Nicolas Williams wrote:
> 
> I thought the DB was 9MB; forgive me for wasting your time then.  If
> it's 17.4MB then the memory usage seems a lot more reasonable.

Daniel, the OP's, database is 9 MB. I don't have his database file, but 
I do have his test code. I used a database of my own that is a similar 
size along with his test code to do my tests.

In my tests sqlite behaves as expected. Daniel is seeing much higher 
memory usage reported from sqlite itself using the same version of 
sqlite, the same test code, and the same OS.

I see a memory usage of about 18 MB for a database copied from a file 
that is 17.4 MB (1K pages) or 12.2 MB (4K pages). I get an expansion 
factor of 1.03 or 1.48.

Daniel is seeing memory usage of 22.2 MB for a database copied from a 
file that is 9 MB. Daniel gets an expansion factor of 2.47. This seems
high.

Since the major difference seems to be the database file we are copying,

I would like to repeat his test with his database file if possible. If 
not possible (perhaps the data is proprietary or personal), then it 
might make sense to see what factors effect this memory expansion ratio.

I was surprised by the magnitude of the change in the size of my 
database file by simply changing the page size. I also tried to change 
the page size used for the memory database, but that had no effect 
(Which is not what I expected, perhaps the page size pragma is ignored 
for memory databases). Changing the cache size reduced the highwater 
memory requirement, but didn't change the memory required to hold the 
database after the copy was completed.

Dennis Cote


_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users

Reply via email to