Maybe to implement audio filters digitally for HDMI audio output?
The implication of this is quite large, as Lisa does not do Audio in the AGA chipset. That's still Paula's task. However, all data that is pushed to Paula via DMA is also flying by Lisa, so we can sniff that and push it into a Paula audio emulation. This gives us a fully digital implementation of the chipset audio. Needless to say that Paula audio emulation is taking up a significant part of the FPGA logic - one explanation for an FPGA that has about twice as many logic cells as the MK2 model had.
Please mind that audio over HDMI is not an advertised feature and it's therefore not covered by official product support. We still have it on the "experimental" features list, as filtering the audio is something that we didn't get right yet. Digital audio is generated with a frequency of 3.58MHz (CCK of the chipset), and re-sampling to 48kHz is something we haven't done before, and you can hear it in the current state of the filters. You do hear a difference of "low pass filter on or off", but there's still a metallic kind-of-noise especially with low-quality samples that is not there on the analogue audio of the computer. We may or may not improve this in the future with a core update.