Speed up S_UltraGrain in Mistika?

I’m a Mistika user and was excited by the release of S_UltraGrain, as there is no real-time grain FX in Mistika currently. We pretty much have to use Cinegrain, which can be cumbersome to set up.

Any way, I have been impressed by S_UltraGrain, but while I get real time playback in Resolve, if I bring the very same 4K clip into Mistika, I only get about 20 fps there.

Is there a Resolve specific optimization that was done on the BorisFX side to get real time playback in Resolve? If so, possible for you guys to have a look if the Mistika experience could be improved?

Single 3090 GPU here, on Windows.