> On Thu, 2 Nov 2000, Patrik Stridvall wrote:
> [...]
> > > * This is a 16bit benchmark but nowadays most applications
> > > are 32bits.
> > > To what extent do these results reflect the performance
> we would get
> > > with a 32bit application?
> >
> > To a very large extent the same I think, especially for
> slow functions.
> > The Win16 <=> Win32 call overhead is not that large.
>
> Maybe (but see Alexandre's post).
It can't be that large some of the
benchmarks are quite OK for Wine.
OK. MoveTo and friend are probably hurt badly by it,
but not BitBlt and friends.
> But it's also possible that 32bit
> applications don't use the same API mix.
True to some extent, see below.
> For instance 16bit APIs might have relied on an API for which we
> performed very poorly. But then when MS introduced the 32bit API they
> added a new 32bit API that is more efficient or more practical to use
> for programmers (can't give an example since I don't know much about
> GDI). Maybe we perform pretty well for the new API. Yet this benchmark
> would lead us to spend all our energy trying to optimize a now mostly
> unused API.
> Ok, it may be far fetched but it's probably something to keep in
> mind...
I think you overestimate the importance of such thing.
Sure it makes some difference but Win16 and Win32 are
not that different. If Wine sucks at Win16 benchmarks
it will probably suck at Win32 ones as well.
