First, I've been working on a Mandelbrot explorer for the past day or two. I used Alphonso's little graphical demo as a base for GDI drawing.


Inside the attachment are my two versions of my code. One nonthreaded, and one that I just tried making threaded.

The threaded one seems to be quite a bit faster, but any time I try to measure the codes using gettickcount(making sure to do one for each thread on the threaded one), it returns roughly the same numbers that the nonthreaded one does. That can't be right.

How would I go about timing this properly?


And please forgive me for the horrible, uncommented, unoptimized code. I just wanted it to work, and didn't care how.


EDIT: Controls are up,down,left,right,z,x,shift,control,and space. Just play around with them.

Where is the call to GetTickCount? It doesn't appear in your source anywhere!

Well, I removed them.

It didn't seem to be working.

So how can we point out the error if you don't post the code that was causing the problem?

Wait a second, I forgot it's threaded. I was adding up the execution time of EACH thread.

So would it be accurate to check the time before I start the threads, and then after they're done?


Attachment on this post has what I think would be more accurate.


I'm not looking for super-accurate timing, just good enough to see the differences between threaded and not.



EDIT: Yeah, this seems about right.

You have hardcoded your threads. What happens if you want to extend it to 8 threads? Will you make 6 more copies of the thread handling code?

Well, I only have a dual-core processor. If I ever get anything better, I'll rewrite the code. Plus, this was mainly made for personal use.