If you have 8-channel audio on the motherboard, I would suggest that you first try using that during the initial proof-of-concept interval of the design development. I am not suggesting that as a solution for the final playback system, but doing that would reduce costs of initial development and could be adequate to measure and listen to the developmental changes in the DSP filters. Depending on the CODEC and implementation (some are much further from horrible than others) you may find that the motherboard audio is adequate for design development up through fine tuning the voicing of the crossover, those latter stage tweaks that mostly balance design compromises in on-axis and off-axis response in the complex in-room response as summed at the listener's ears and as affected by the nonlinear nature of human hearing and perception, psychophysiology and psychoacoustics.

Before completing design development you may choose to alter the design to something very different from the initial concept, and that may require more than 8 channels, more especially if you decide to add multiple individually equalized subwoofers (as I would suggest) to better distribute and adjust sources of excitation of the modal response of the listening room. Modal response tends to dominate the problem around and below the Schroeder frequency, constructive and destructive interference of Eigentone fundamentals and low order harmonics associated with listening room modal resonances (standing waves associated with room boundary geometry), and likewise perhaps also interfence with a low frequency Helmholtz resonance associated with vents and openings connecting the listening room air mass compliance to other spaces.