> unfortunately the machines in the device are both slower and running a lot of processes which causes the software timer to be slow by about 8 seconds per hour.
But you were using QueryPerformanceCounter?
Also bear in mind that QPC could give some very odd results if the processor is variably clocked (as in a laptop) to save power.
What exactly is your embedded device?
If it's running processes, then that implies a scheduler of some sort as well. Unless the scheduler is entirely cooperative, having a hardware timer interrupt is all but essential.
Having found the hardware timer, is that stable for your needs?
Make sure you wear insulated gloves for this Dr. Frankenstien -- someone might get hurt.
Originally Posted by cyberfish
That's quite a good suggestion. We're actually running servers with Quad-core opterons, and after a bit of reading it looks like AMD implemented "Independent Dynamic Core Technology" which clocks each core independently. I'm going to check out if disabling that helps anything. Thanks!
Originally Posted by Salem
As for the hardware timer. I'm now trying to devise a software side solution for syncing to a signal over RS-232. The existing implementation is to use ReceiveFrom() to sync via UDP without constantly polling. Anybody know of a similar functionality for RS-232?
I'm guessing you've got an OS on top of that lot?
So which is it?