I don't know how this might apply to MorphOS but I'd like to think there is some way it could boost the user experience with Linux -

What if, instead of swap partition virtual memory, a largish chunk of graphics card memory could be used for this purpose?

It would probably be necessary to balance OS access to the graphics card memory to prevent impacting video playback but wouldn't this still be faster than hard drive access? Would this come close to being a work-around for the Efikas fixed system memory size?

This has been discussed a lot on other parts of the forum. The basic problem is that you have to give up hardware acceleration of 2D and 3D, using high resolutions or storing graphics data in video RAM - or make changes to the kernel DRM and X.org driver modules at least to only use a certain amount of memory, and never to disable the card or lock it unless it's atomic.

Most of the more advanced accelerations just wouldn't work. The XGI DRM looks fairly simple, so it might, but we're going to ship 64MB cards here. 32MB of swap memory isn't much. Most 256MB+ cards will not work in an Efika.

In the end it's basically just not worth the time or the hassle. Just buy a hard disk..