Gil, as the author of the program in question I may be biased. But in all honesty the simulator should be 100% accurate in measuring the timing of code blocks as long as the code blocks do not interact with outside hardware nor use the serial port.

The execution time the simulator reports for code execution of completely internal code should be 100% accurate. It is quite simple for the simulator to calculate exection time since the clock cycles for each instruction is known.

If you find any discrepancy please let me know since that would be indicative of a bug.

This is not to contradict one of Erik's postings where he cautions against the use of simulators. Simulators do have limitations as he points out. However, I find simulation to be very very useful when testing program *logic*. An emulator simply isn't necessary to determine whether your floating point routines are working. An emulator is more useful, of course, when interacting with external hardware. As always, use the right tool for the job.

Craig Steiner