[waydabber]

There is some difference, but it is marginal indeed at least in terms of speed. Via the traitional route with a real dummy, the display hardware needs to produce two sets of displayport output streams (one of which is converted to HDMI via a DisplayPort-HDMI controller chip MCDP2900 to drive the Dummy) + sync up the two displays in terms of vertical sync (this does not always work well, this is why real HDMI dummy users experience mouse jittery sometimes) and also has to scale the full-res framebuffer to two independent display (but scaling is done super efficiently on M1). With BetterDummy all this is not needed obviously.

For ppl who until now had to resort to mirroring the internal displays (as M1 MacBooks - before the new MBPs - support only the internal display + a single external display) the benefit is more obvious, they now can use clamshell mode and don't have to drive the full MacBook display hardware (with brightness turned down to zero) all the time.

[bluesign]

Thanks for the detailed answer. Btw any chance this can work on virtualized environment without real GPU. Last time I gave this a shot (virtual screen), couldn’t manage to get it work with hidpi.

[waydabber]

This is an interesting question, honestly I have no idea. It might be that these APIs are relying on the presence of a GPU for acceleration (as the private framework APIs used are made by apple for Sidecar and AirPlay primarily). If you test this, please let me know about the results at the project GitHub page (https://github.com/waydabber/BetterDummy). Thank you!