"cgfourteen
[...]
The 8Mb VSIMM allows a resolution of 1280x1024 at
a depth of 24
bits;
In my experience it allows much better than that; my SS20 + 8MB cg14
supports 24bpp at 1600x1280 and at 1920x1080. (Now if I could just
find a(n affordable) flatscreen that does either of those....)
Does anyone
know if the S24 (SS5 AFX framebuffer) was accelerated?
I didn't find this
mentioned on the above URL, but perhaps if you
look at it you'll recognize it in the list.
It is accelerated as compared to a dumb memory-mapped framebuffer, but
not very much. It has hardware support that improves filling of
rectangles with what X calls FillSolid and FillStippled (but not
FillTiled or FillOpaqueStippled, except to the extent that they can be
simulated using FillSolid and FillStippled). It also accelerates
screen-to-screen copies.
In neither case does it accelerate it very much. Rectangle fills are
accelerated by a factor of approximately 16 (one 64-bit write fills 32
pixels instead of 2 pixels - I say "approximately" because the exact
value depends on the width of the rectangle and how it is aligned).
Screen-to-screen copies are accelerated such that copying 32 pixels
needs just one (64-bit) write, rather than 16 reads and 16 writes. See
ftp.rodents.montreal.qc.ca:/mouse/docs/Sun/S24/memory-map if you want
details.
Useful, perhaps, but not nearly as nice as the cg6, which can fill or
move a whole rectangle up to the screen size in...umm...best case, I
think, eight writes - more like 15-20 if you don't know the state of a
few pieces of the card.
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
