>        Well I've been fooling around with various TC FB's (PMAGB-B,
> PMAG-C, PMAG-E, PMAG-DV) and I have a proposition or two to make.
>
>        1.) Scrolling on the PMAGB-B and PMAG-DV are SLooowwwwwww, and it
>                stops the entire computer when it does. (I'm talking about
>                in the system console, not in X)

- Because the graphic rendering is done within highly blocked interrupt
  context.
- To fill screen with small gryphs is generally bad for performance.
- rasfont render needs improvements to squeese off the maximum CPU
  and memory bandwidth.
- Half rated 12.5MHz TC is noticeably slower for framebuffer rendering.

>                Can we make like Ultrix
>                and shift the display up about five lines at a time when
>                it reaches the bottom of the screen instead of just one? 

(N)WSCONS once did it.  Current code base, however, does
scrollup-per-line because many complained the behaviour.  It can be
switched back when compiled so.

SFB hardware accelerated WSCONS video console has been available on
DEC3000.  People expressed satisfaction with the supreme scroll
performance. 

One of NetBSD/pmax long standing TODOs is WSCONS transition.  All of
framebuffer drivers were written a few years ago (See dev/tc/
directory), but the progress of keyboard/mouse driver has stucked for
long time.

Tohru Nishimura