27 May
2014
27 May
'14
4:26 p.m.
In article <CADBZjLYMcZZDLa+CG8dxHWj5R35_Yn22VZW6eNwbSDZi6j7O9A at mail.gmail.com>,
Josh Dersch <derschjo at gmail.com> writes:
The Imlac PDS-1 (late '69, early 1970) (and later
models) also fall into
the GPU model, each PDS-1 contained both a general purpose 16-bit processor
(similar to the PDP-8 only 16 bits wide) and a Display Processor, dedicated
to processing a display program (slightly more advanced than a display list
but not much). They both shared the same memory, so a very complicated
display slowed the main processor quite significantly :).
Yeah, I think from the late 60s to mid 70s this was a very common
architecture for low-end graphics workstations/terminals that used
vector refresh displays.
It's not until you get to cheaper memories and people switch to raster
graphics that you get what we think of now as a "classic 80s workstation"
where the "GPU" once again gets stupid and simply performs video
scanout/refresh of a frame buffer and the CPU does all the "graphics".
High end workstations have real GPUs that do rasterization of
geometric primitives and/or pixel operations (bitblt, etc.) that act
as coprocessors to the CPU. They may share a bus with the CPU
(slower) or have their own private bus (faster) to access display
memory. Once the display memory becomes separated from the CPU's
address space, the whole business of coordinating them gets more
complicated.
--
"The Direct3D Graphics Pipeline" free book
<http://tinyurl.com/d3d-pipeline>
The Computer Graphics Museum <http://ComputerGraphicsMuseum.org>
The Terminals Wiki <http://terminals.classiccmp.org>
Legalize Adulthood! (my blog) <http://LegalizeAdulthood.wordpress.com>