First note I'd make is to NOT program your solution as a progression of mathematical statements written on a sheet of paper. Use algebra to combine everything possible before hand and shape the equation such that you perform only one division, preferrably the last operation. The first operations in the process should be the lossless calculations, the last should be the lossy.

Don't consider mathematics as being in binary, converting to hex and converting to decimal... its all unnecessary. You keep your algorithm in free binary and you can keep all your fractional carries (though reasonable roundoffs are necessary) until you have your final answer. ONLY THEN, convert to a display from directly. Tabular systems with linear interpolation can be quite good.

The valid point has been made that displaying RPM in many digits is problematic, particularly if your update rate turns the numbers into a blur of Numeric LEDs looking like 888s (think about that).

Its better to provide customers with information that is useful to them rather than the best calculation for an instant in time that has already passed.

For RPM, I'd suggest initial display accuracy roundoff to the nearest 25RPM.
You can add additional code should you choose such that once the device has detected RPM relatively stable, that it switch to a 5 or 10 RPM display roundoff. This may be appropriate for mechanics who are tuning for a target RPM.

Honestly... to have the low digits displaying odd digits would be quite distracting. Think of the useful display numbers as being similar to our choices in monetary coinage. We COULD all carry pennies around for the ultimate accuracy in exchanges, but we prefer loading quarters into cola machines even if the roundoff costs us a few more cents.

A comment on customer psychology in displays. Its better to give them "[2500]" in LEDs than "[2.5] x1000".
Partial information annoys customers and may lead them to a competitor. Its WORTH including those extra numeric LEDs EVEN IF they ALWAYS display "[00]".

Its not enough to be a great engineer to succeed... you have to be part businessman too.

aka j