Hi, So the VC++ 7.1 dll is much faster than the VC++ 6, up to 20% in some operations, and the "MS mixed dll load bug" is not a problem, assuming sqlite doesn't mix native code with managed .net code :).
Anyone using other compiler to build sqlite on Windows ? Maybe Borland ? I seems there is a lot o room for faster sqlite.dll by using other compiers than VC++ 6. We can run some tests to see which is faster ... Ionut Filip -----Original Message----- From: Robert Simpson [mailto:[EMAIL PROTECTED] Sent: Saturday, February 26, 2005 1:06 AM To: sqlite-users@sqlite.org; 'Keith Herold' Subject: RE: [sqlite] sqlite performance variationin linux and windows > -----Original Message----- > From: Keith Herold [mailto:[EMAIL PROTECTED] > Sent: Friday, February 25, 2005 3:57 PM > To: sqlite-users@sqlite.org > Subject: Re: [sqlite] sqlite performance variationin linux and windows > > http://www.adtmag.com/article.asp?id=7421 > > http://msdn.microsoft.com/library/default.asp?url=/library/en- > us/dv_vstechart/html/vcconMixedDLLLoadingProblem.asp > > It's been awhile since Iooked at this, but I distinctly remember > reading from one of the product managers at MS that this wouldn't be > fixed until 2005. The basic details were, I think, that the 'native' > multithreaded C++ applications, under heavy load, would kill deadlock > and halt. Since I work for a speech recognition company, the cpu is > *always* under heavy load, and the problem has kept us from moving to > .NET . Otherwise, I think we would probably do switch, since there > are some frustrating bugs in VC6 that have been fixed. > > I also don't know if this has been fixed already, or not. According to the article on MSDN's website this only pertains to mixed-mode C++ applications. In otherwords, a C++ application containing both C++ managed and unmanaged code. Code written entirely managed or entirely unmanaged (native -- such as SQLite) are unaffected. Robert
