Lpc2000

I'm trying to squeeze a few more cycles out of one my LPC, and I'm
looking for an optimized way to divide by 10 and leave the quotient and
remainder.  I.e., a faster way to do this:

#define DIVIDE_BY_10(D, Q, R) Q=D/10; R=D%10;  // D is unsigned long

[It's to support a prehistoric data collection system, writing LOTS of
BCD ASCII data to a SRAM cartridge.]

Any info would be appreciated.

Thanks.