Update: I am running on a hunch, then when you use DEVOUT to print integers to a screen in something like a 1 to 300 count, it is that printing to the screen that is holding everything up... and cl52e is winning this race. I will try to prove it with a hello world example to see if this theory is correct...
But making it display to modulus 50 instead of every single integer improved the speed by about 3 times in this test on Win98. On Mon, Mar 29, 2010 at 3:58 PM, smu johnson <smujohn...@gmail.com> wrote: > Hi, > > We have a little benchmarking section in a Clipper program we wrote that > gives a percentage of how fast it's running overall. I believe it is doing > 500 SIXCDX row creates etc... > > For Win98 programs, are 32-bit .EXE's compiled with Harbour + TDM Mingw > have around 25% on that list, whereas when they are around 200%+ when > compiled as 16-bit with cl52e. > > This problem doesn't happen with XP and above. > > So, I'm wondering if for Win98, the Harbour-made .EXEs... have to emulate > some 32-bit mode? Or maybe the processors in most Win98 machines are 16-bit > CPUs? I really have no idea... I wish I knew more about this... but in a > nutshell, our test is running WAY slower than cl52e + Win98. > > Thank you for reading. > > -- > smu johnson <smujohn...@gmail.com> > > -- smu johnson <smujohn...@gmail.com>
_______________________________________________ Harbour mailing list (attachment size limit: 40KB) Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
