I think ( maybe I am wrong) what Sriharsha wants to know is that, if there is any software method of knowing whether a X MHz crystal is really kicking at X MHz ?

I dont suppose that there is any such software method.

Even I have had this predicament in a code that I did for calibrating a MCU based programmable timer. And when I wanted to "certify" the accuracy levels to the client, I finally had to resort to a high end Frequency counter from HP that was borrowed for the purpose. We found the actual deviation was small ( less than 0.5%) and thus were able manage the situation with a particular crystal that was closest. ( OK no effort was made to correct the temperature drift issues - but then the application could make do with what we achieved.)

But as you mentioned the method deployed will to a lerge extent depend on the "accuracy" AND "precision" that you want to achieve finally.

Raghu