What differentiates a Workstation from a PC (Re: What should GNU run on (was Re: what kinds of things . . .))

[Chris Calabrese[mav]]

In article <5687 at ficc.uu.net>, peter at ficc.uu.net (Peter da Silva) writes:
> In article <1528 at convex.UUCP>, datri at concave.uucp (Anthony A. Datri) writes:
> > Personally, I find it difficult to call something running MS-DOS with
> > a fairly low-res screen a workstation.
> 
> How about something runing UNIX and X-windows with a 1K by 1K display?
> That's what you'll find in high-end PCs these days.

I keep seeing references to graphics capabilities and to raw cpu power
in this discussion, but I think people are overlooking one important
aspect of the PC vs Workstations wars - the technology used to
implement the display hardware.

Most PC display adapter boards are simple memory maps between board
hardware and the pixels on the screen.  The main cpu is responsible
for things such as bit-blitting, off-screen image copies (for things
like fast popup menus), and color rectangle fills.

In "workstations", either the graphics board can do many of these
things itself (i.e. the Sun color framebuffers or the Silicon Graphics
machines), or the workstation is only used for the display and keyboard
functions (i.e. X server only workstations or the AT&T 630) and
another machine is used for compute bound processing.  In either of
these events, the user level computation is seperated from the graphics
level computation.

A 25Mhz '386 AT&T box running X windows is slower than
a Sun 3/60 running NeWS, yet the '386 has more raw cpu power.  I
believe this is due to the above mentioned differences in the way the
graphics boards work.
-- 
Name:			Christopher J. Calabrese
Brain loaned to:	AT&T Bell Laboratories, Murray Hill, NJ
att!ulysses!cjc		cjc at ulysses.att.com
Obligatory Quote:	``Anyone who would tell you that would also try and sell you the Brooklyn Bridge.''