What is the difference between parallel port control board and mesa control boar

Both systems are the same in the servo thread. All the same modules work together to create a new waypoint for each axis every 1ms.

In the case of the parallel port this waypoint is sent to code running in the base thread (typically 20us +/- 10us), which drives the parallel port pins.
With a Mesa card the waypoint is sent to an FPGA (base clock period 20nS +/- 1nS) which creates the step pulses.

Both can work perfectly adequately. But the Mesa card can step much faster and at much more accurate frequencies, which can sometimes allow for higher motor speeds even with the same hardware.

Also, the p-port has a limited number of pins, whereas Mesa hardware can be expanded to thousands of pins if needed.

The p-port is only really useful for step/dir and low-count encoders. Mesa cards can drive a much wider variety of hardware (analog servo, resolver....)

The p-port is cheaper.

I would suggest starting with a parallel port if you don't know what you need.

tuzki wrote:

As you said, if you use the mesa control board, LinuxCNC will send the waypoint to the mesa control board for processing, so why the speed of the mesa control board is faster than that of the parallel port control board.

The limitation with software step generation is the rate at which the realtime system can service the realtime threads. With a typical latency of about 10uS you can't really call that thread more frequently than every 20uS.