Data comming in on the parallel port is read 8 bits at a time. The easiest ways would be to break up the 8 bits by performing a shift on the 8 data. This would allow you to take samples from 8 chips at once. The software routines to do this would actually be very fast depending on the compiler. Doesn't the ADC0838 require a clock signal as well as the data? If it does you will need to clock the chip before reading each bit.