I thought those were very interesting figures - showing memory
allocation is slower and I would have expected possibly more overhead
on calling a function (although not necessarily - user functions were
always virtual).
However, I would like to emphasise this is a highly irrelevant test
for most RB users - you need to be allocating and calling functions
on TEN MILLION objects to notice a few seconds difference!
The emails didn't say anything about the test protocols - for
anything memory intensive like this you need to ensure the
environment is as close to identical as possible. This means, at a
bare minimum, booting the machine from scratch each time and ensuring
nothing else is running.
it may sound obvious but also ensure you test compiled binaries and
not run things from inside the IDE because then you're testing
debugger overhead :-)
Far more useful, would be tests that
- iterate across simple data
- perform complex calculations (a true test of the compiler)
- exercise the drawing logic
- exercise the database
- load controls such as listboxes, strings into edit fields... As
Dazz pointed out, listbox population in RB2006 is much faster than
previously.
Sorry I'm too busy to create such tests myself but I encourage those
with time to carry on with this endeavour :-)
I *am* glad to see people turning their attention to the raw speed
comparison of RB versions, rather than taking the IDE as an
indication that their own programs will necessarily be slow (unless
they are writing an IDE).
Andy
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
