I have my own issues with code speed. Some techniques that can help:
1. Update analog and digital inputs / outputs only on a change-of-state, if possible.
2. And / or gate the inputs / outputs with as slow of a RTC (Real-Time-Clock) as will keep the process functioning.
3. If you have to process huge arrays of data at high-speed and at a repetitive high rate, a standard pc or even some low-end workstations may not have the raw horsepower required.
I also have a question pertaining to "tweeking" for speed.
What kind of penalty (in CPU cycles) is exacted for using a global variable as opposed to a local variable?
I design, build and program our in-house test equipment to test the 3-phase surge protection devices we sell.
Some of them are elaborate test sequencers that supply constant-current in steps of 100uA to 100mA and 24vdc to 3kvdc.
This kind of equipment has an lengthy init / POST sequence and needs to run auto-calibration and auto-diagnostic routines in the background when the user is not testing a product. So I build hybrid task managers of 4D I64 arrays that allow for both preemptive and some cooperative multi-tasking. The global variables are easy to access from anywhere with special reset/init/read/write/edit/update vi's).
The question becomes: Does using a global variable(4D I64) as a task registry have a serious code speed penalty over the same size array made as a local variable? As a local variable it requires a lot of links and handles to access and edit its content. Its dificult to tell if the extra address translation required by using global variables is more of a penalty than having lots of handles and array handling vi's to work with large local variables.
NOTE: Only a single "book" and "page" of the task registry is active at a time. The task mgr only scans the contents of 1 page.
