Quote from: SiegeBreaker on December 12, 2012, 03:16:22 AM

Quote from: ckolivas on December 11, 2012, 12:18:27 AM

...
cgminer times how long it takes for the gpu code to "return" after it's been given work. Unfortunately with windows, the timer resolution is so shithouse at 15ms that I have to sample many many iterations of the GPU code and then average them to see how long it took.
...

cklovias, there are several ways to overcome this drawback of the simple windows time functions. Probably the two easiest ways are using the code from this explanation http://msdn.microsoft.com/en-us/magazine/cc163996.aspx, or by using the .Net stopwatch class (for your code, you would have to call into a c++/cli dll wrapper around the class, then you have to deal with people needing the .Net framework installed too...maybe make it optional?)

I like using dynamic on my main windows computer, but I have been sticking with 2.7.5 because it has better dynamic behaviour for me. I have just been thinking, 'oh well, make it until the ASICs come out and that's good enough!'

I've already found issues with USB and the fact that the clock is so unreliable on windows (XP).
This is where I've found more useful info (than msdn) to solve the problem - since MS themselves really don't seem to care, it seems pointless looking at msdn Tongue

http://www.geisswerks.com/ryan/FAQS/timing.html