I agree with Steve. If you prescale the 200 MHz down to a period that the 8051 can measure there is NO NADA NONE loss of accuracy due to the divider. The errors you will see will be two fold. One is due to the stability and accuracy of the 8051 crystal. The second is the quantization error due to the 8051 gating being async to the divided down 200 MHz. This latter error is nominally 1 sample clock per period measured but could be 2 I suppose. If, as I said before, the quantization error is too high then just divide the 200 MHz down some more and measure a longer period.

Michael Karas