27 May
2014
27 May
'14
2:02 p.m.
In article <5384E178.9020408 at sydex.com>,
Chuck Guzis <cclist at sydex.com> writes:
On 05/27/2014 10:08 AM, Richard wrote:
I'm not sure buffered is the right term to use here. Everything is
stored in a buffer unless you use a storage tube display or hardcopy
output. The more typical term for what you describe above is "display
list" for vector displays and the term even lives on in the OpenGL
API. Once people switched to raster displays, then the term more
commonly became "frame buffer" to refer to the raster memory and
"display list" to refer to a buffer of geometry that was fed to a
rasterization pipeline.
See "On the Design of Display Processors", T. H. Myer and I. E.
Sutherland, Communications of the ACM, June, 1968
<http://cva.stanford.edu/classes/cs99s/papers/myer-sutherland-design-of-display-processors.pdf>
The "wheel of reincarnation" refers to an endless trip around a circle
between a "dumb frame store" where all the display maintenance is
handled by the CPU as you describe above and an "intelligent graphics
controller" (i.e. GPU) that handles all the display maintenance and
appears as a memory store to the CPU.
The earliest example I can think of is in the SAGE system where the
frame store was literally one (or more?) band on drum memory. The
drum memory cycle time was synchronized to the display refresh rate.
You read instructions off the drum and they literally fed the refresh
circuitry. This is dimly from memory, but that puts the display list
in the SAGE system in the late 1950s>
Early GPUs were basically a dedicated CPU to refreshing the display.
Examples are the GT40, where the GPU controller is an 11/05, or
something like the Megatek 6014 where the GPU controller is a Data
General Nova 3.
<http://en.wikipedia.org/wiki/DEC_GT40>
<http://terminals.classiccmp.org/wiki/index.php/Megatek_Megraphic_6014>
Interestingly enough the GT40 has the 11/05 to handle the management of
the display list and handle communications processing and the VT11 board
set handles refresh of the display from the display list. You could stick
a VT11 board set in any UNIBUS PDP-11 and have it manage the display by
refreshing from the display list memory. A GT40 offloads display list
refresh from the main CPU bus and offloads the main CPU from handling
communication to/from the graphics display.
Some points on a timeline:
1958 SAGE display system [4]
1969 Evans & Sutherland LDS-1 [1]
1973 DEC VT11 [2]
1977 HP 1350A Graphics Translator [3]
[1] <http://en.wikipedia.org/wiki/LDS-1_(Line_Drawing_System-1)>
[2] <http://manx.classiccmp.org/search.php?cp=1&q=VT11>
[3] <http://www.hpmuseum.net/display_item.php?hw=169>
[4]
<http://bitsavers.trailing-edge.com/pdf/ibm/sage/3-62-0_Display_System_Vol1_Aug58.pdf>
--
"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>
It's a vector display, just without arbitrary
graphics.
A point of curiosity--who had the first *buffered* graphics display?
That is, where the display contents were continuously refreshed from a
memory buffer. In the old days, you pretty much had to continuously
redraw a display, unless you were using something like a storage tube.
The more complex the display, the more flicker involved.