"Whats the best way of measuring execuation time of a program with keil c."

If possible, run your code in the simulator & use the Performance Analyser.

If you need to measure it in the target, then an ICE would be best.
Failing that, toggle a port pin at suitable points in the code and use a scope or something to measure the interval.

"I used to measure it with the simulator using the counters..."

Why? the Performance Analyser gives you this for free!

"how do i know that my overall loop is less than 1msec ?"

Your processor datasheet will tell you how long each instruction takes to execute - just add 'em up!