Currently VBA-M's joybus code intentionally uses blocking sockets to ensure that there is no loss of sync. This can cause major slowdowns before the bios has started to use the joybus connection, as many people undoubtably noticed. It will probably be fixed/improved in the future, however I don't really feel like working on this until either:
gba ucode is HLE'd
dsp lle gets some sort of recompiler
rest assured either (or both) of these things will happen - but like always, no timelines.
gba ucode is HLE'd
dsp lle gets some sort of recompiler
rest assured either (or both) of these things will happen - but like always, no timelines.