4 ms·

The history of the modern graphics processor

59 points by kaptain 12y ago

aw3c2 12y ago
An article like that should really cite it's sources.
- Svenstaro 12y ago
  its
blt 12y ago
Ha ha, look at the capacitors soldered across the through-hole ICs on that first ATI board from 1987. Only took a couple years to get to sophisticated SMT boards.
Lennu 12y ago
It always amazes me how so bad quality graphics back in the 90s or early 2000 looked so bad but still everybody thought they were great and amazing (at least I thought). Probably we think now that we have good looking graphics but after 15 years of development they aren't so good anymore.. AMD/ATI and NVIDIA have done great work over the years.
- bluedino 12y ago
  Typically the things that previous made graphics 'good' were simply resolution and color depth. Early 8-bit home computers were often limited to just 1 or 2 colors per sprite. Machines such as the NES increased this to 3 or 4 colors per sprite, which makes a dramatic difference. Along came something like the SNES, which although it shares the same resolution as the NES (256x220 or something along those lines), it had access to 32,000 colors instead of the 64 available to the NES, and it could use up to 16 per sprite. That's the main reason why SNES screenshots look so much better then SNES screenshots. Once 3D came along, graphics looking 'good' became more complicated than just resolution and color depth. The original Playstation suffered from texture maps that showed artifacts, became distorted, and had edges that appeared to 'jump' around. As the hardware became faster and fixed those issues, more techniques were added for lighting, shading, and a hundred other enhancement which got us to where we are today.
- higherpurpose 12y ago
  Mainly because you were comparing the new graphics with the old ones, and marveled at the "progress", even if they were nothing like real-life. Also, around the time when the switch to 3D graphics happened, the 3D ones looked a lot uglier than some 2D games that seemed prettier, which had better art and colors.
- boomlinde 12y ago
  I think it's less of a matter of good/bad than a bunch of other factors such as detail, immersion and novelty compared to what was available before. The graphics looked good in the context of earlier graphics systems, and I doubt it will be different with the current generation of 3D games.
rayiner 12y ago
> The evolution of the modern graphics processor begins with the introduction of the first 3D add-in cards in 1995, followed by the widespread adoption of the 32-bit operating systems and the affordable personal computer. What? No. The evolution of the modern graphics processor begins with the development of commercial 3D systems in the 1980's. Jim Clark founded SGI in 1982 based on his work into hardware acceleration of geometry computations for 3D at Stanford. By the mid-1980's, SGI workstations were able to handle 3D modeling and animation locally: http://en.wikipedia.org/wiki/Silicon_Graphics#IRIS_2000_and_3000_series http://en.wikipedia.org/wiki/Silicon_Graphics#IRIS_2000_and_.... You can find some of Clark's papers on the hardware here: http://www.computer.org/csdl/mags/co/1980/07/01653711.pdf http://www.computer.org/csdl/mags/co/1980/07/01653711.pdf (1980); http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.359.8519 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.359.... (1982). I can't find a good reference describing the rasterization side of the pipeline, though the second paper describes it very briefly on page 132. There's a really neat website going into details of early 3D consumer chips: http://vintage3d.org http://vintage3d.org. It's very interesting, though again it should be noted that these products came out in the 1990's, well after hardware acceleration of 3D was well-established in the workstation market.
- adventured 12y ago
  Article has tons of holes in it. They didn't even mention Pixar once: http://en.wikipedia.org/wiki/Pixar_Image_Computer http://en.wikipedia.org/wiki/Pixar_Image_Computer
  - gaius 12y ago
    Or http://en.wikipedia.org/wiki/Original_Amiga_chipset http://en.wikipedia.org/wiki/Original_Amiga_chipset
- greggman 12y ago
  I could be wrong but didn't Evans and Sutherland have 3D accelerated graphics in the 70s for military flight simulators? Or is my memory bad?
jasode 12y ago
A nice complementary text (programmer's perspective) to the history of the hardware is the epic stackexchange answer: http://programmers.stackexchange.com/a/88055 http://programmers.stackexchange.com/a/88055
- ARothfusz 12y ago
  Very nice. Though I wonder why the author thought we (3Dfx) were "strongly against OpenGL"? One of our founders, Gary Tarolli, worked on Iris GL[1][2]. We weren't against it, but our pixel pipeline was in the wrong order to implement it precisely. Plus there was the GL tradition of handling things in software that your hardware didn't implement, and we hated the "slow path". Ok, maybe we were against it a bit :-) Coming back to me now. [1] http://tech-insider.org/unix/research/1990/1211.html http://tech-insider.org/unix/research/1990/1211.html [2] http://techpubs.sgi.com/library/manuals/1000/007-1210-060/pdf/007-1210-060.pdf http://techpubs.sgi.com/library/manuals/1000/007-1210-060/pd...
mpweiher 12y ago
More wrong: the 6845 was not used in the Apple II.
- joezydeco 12y ago
  Correct. It was used in some of the 80-column addon cards, but Woz drove the video signal directly from a sea of TTL gates. It was pretty amazing stuff for the time.
adwf 12y ago
It's an interesting topic in that I remember just how much of a revolution the original Voodoo card was to gaming (for me anyway). These new cards didn't just enable slightly prettier graphics like a new card would today, they enabled entire new genres of game. If anything, the art quality actually took a step back for a while (some of those sprites were gorgeous), but the new 3D effects more than made up for it. I imagine products like the Oculus Rift will hopefully have the same effect on gaming. New ideas that we could barely imagine before will now be within reach.
- dualogy 12y ago
  > they enabled entire new genres of game Such as? There have been great enjoyable DOS-based 3D games for a number of years ever since Wolfenstein so I'm curious what 'new genres' you're thinking of that the introduction of a 3D graphics processor enabled.. ;)
  - penguindev 12y ago
    Right. Even Unreal had software rendering (and looked damn good - water, sky, fog).
stox 12y ago
Some interesting history here in a discussion of 3DFX: https://www.youtube.com/watch?v=3MghYhf-GhU https://www.youtube.com/watch?v=3MghYhf-GhU
- purplequark 12y ago
  If you go to http://www.computerhistory.org/collections/oralhistories/ http://www.computerhistory.org/collections/oralhistories/ , and click on the Videos tab, you can view it with a sync'd/searchable transcript. It's a beta feature right now, and I know they'd appreciate feedback. Incidentally, while only a few of the oral histories have videos up, there's 500 or so transcripts up, will worth a perusal.

59 points by kaptain 12y ago

aw3c2 12y ago

An article like that should really cite it's sources.

Svenstaro 12y ago
its

blt 12y ago

Ha ha, look at the capacitors soldered across the through-hole ICs on that first ATI board from 1987. Only took a couple years to get to sophisticated SMT boards.

Lennu 12y ago

It always amazes me how so bad quality graphics back in the 90s or early 2000 looked so bad but still everybody thought they were great and amazing (at least I thought). Probably we think now that we have good looking graphics but after 15 years of development they aren't so good anymore.. AMD/ATI and NVIDIA have done great work over the years.

bluedino 12y ago
Typically the things that previous made graphics 'good' were simply resolution and color depth. Early 8-bit home computers were often limited to just 1 or 2 colors per sprite. Machines such as the NES increased this to 3 or 4 colors per sprite, which makes a dramatic difference. Along came something like the SNES, which although it shares the same resolution as the NES (256x220 or something along those lines), it had access to 32,000 colors instead of the 64 available to the NES, and it could use up to 16 per sprite. That's the main reason why SNES screenshots look so much better then SNES screenshots. Once 3D came along, graphics looking 'good' became more complicated than just resolution and color depth. The original Playstation suffered from texture maps that showed artifacts, became distorted, and had edges that appeared to 'jump' around. As the hardware became faster and fixed those issues, more techniques were added for lighting, shading, and a hundred other enhancement which got us to where we are today.
higherpurpose 12y ago
Mainly because you were comparing the new graphics with the old ones, and marveled at the "progress", even if they were nothing like real-life. Also, around the time when the switch to 3D graphics happened, the 3D ones looked a lot uglier than some 2D games that seemed prettier, which had better art and colors.
boomlinde 12y ago
I think it's less of a matter of good/bad than a bunch of other factors such as detail, immersion and novelty compared to what was available before. The graphics looked good in the context of earlier graphics systems, and I doubt it will be different with the current generation of 3D games.

rayiner 12y ago

> The evolution of the modern graphics processor begins with the introduction of the first 3D add-in cards in 1995, followed by the widespread adoption of the 32-bit operating systems and the affordable personal computer. What? No. The evolution of the modern graphics processor begins with the development of commercial 3D systems in the 1980's. Jim Clark founded SGI in 1982 based on his work into hardware acceleration of geometry computations for 3D at Stanford. By the mid-1980's, SGI workstations were able to handle 3D modeling and animation locally: http://en.wikipedia.org/wiki/Silicon_Graphics#IRIS_2000_and_3000_series http://en.wikipedia.org/wiki/Silicon_Graphics#IRIS_2000_and_.... You can find some of Clark's papers on the hardware here: http://www.computer.org/csdl/mags/co/1980/07/01653711.pdf http://www.computer.org/csdl/mags/co/1980/07/01653711.pdf (1980); http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.359.8519 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.359.... (1982). I can't find a good reference describing the rasterization side of the pipeline, though the second paper describes it very briefly on page 132. There's a really neat website going into details of early 3D consumer chips: http://vintage3d.org http://vintage3d.org. It's very interesting, though again it should be noted that these products came out in the 1990's, well after hardware acceleration of 3D was well-established in the workstation market.

adventured 12y ago
Article has tons of holes in it. They didn't even mention Pixar once: http://en.wikipedia.org/wiki/Pixar_Image_Computer http://en.wikipedia.org/wiki/Pixar_Image_Computer
- gaius 12y ago
  Or http://en.wikipedia.org/wiki/Original_Amiga_chipset http://en.wikipedia.org/wiki/Original_Amiga_chipset
greggman 12y ago
I could be wrong but didn't Evans and Sutherland have 3D accelerated graphics in the 70s for military flight simulators? Or is my memory bad?

jasode 12y ago

A nice complementary text (programmer's perspective) to the history of the hardware is the epic stackexchange answer: http://programmers.stackexchange.com/a/88055 http://programmers.stackexchange.com/a/88055

ARothfusz 12y ago
Very nice. Though I wonder why the author thought we (3Dfx) were "strongly against OpenGL"? One of our founders, Gary Tarolli, worked on Iris GL[1][2]. We weren't against it, but our pixel pipeline was in the wrong order to implement it precisely. Plus there was the GL tradition of handling things in software that your hardware didn't implement, and we hated the "slow path". Ok, maybe we were against it a bit :-) Coming back to me now. [1] http://tech-insider.org/unix/research/1990/1211.html http://tech-insider.org/unix/research/1990/1211.html [2] http://techpubs.sgi.com/library/manuals/1000/007-1210-060/pdf/007-1210-060.pdf http://techpubs.sgi.com/library/manuals/1000/007-1210-060/pd...

mpweiher 12y ago

More wrong: the 6845 was not used in the Apple II.

joezydeco 12y ago
Correct. It was used in some of the 80-column addon cards, but Woz drove the video signal directly from a sea of TTL gates. It was pretty amazing stuff for the time.

adwf 12y ago

It's an interesting topic in that I remember just how much of a revolution the original Voodoo card was to gaming (for me anyway). These new cards didn't just enable slightly prettier graphics like a new card would today, they enabled entire new genres of game. If anything, the art quality actually took a step back for a while (some of those sprites were gorgeous), but the new 3D effects more than made up for it. I imagine products like the Oculus Rift will hopefully have the same effect on gaming. New ideas that we could barely imagine before will now be within reach.

dualogy 12y ago
> they enabled entire new genres of game Such as? There have been great enjoyable DOS-based 3D games for a number of years ever since Wolfenstein so I'm curious what 'new genres' you're thinking of that the introduction of a 3D graphics processor enabled.. ;)
- penguindev 12y ago
  Right. Even Unreal had software rendering (and looked damn good - water, sky, fog).

stox 12y ago

Some interesting history here in a discussion of 3DFX: https://www.youtube.com/watch?v=3MghYhf-GhU https://www.youtube.com/watch?v=3MghYhf-GhU

purplequark 12y ago
If you go to http://www.computerhistory.org/collections/oralhistories/ http://www.computerhistory.org/collections/oralhistories/ , and click on the Videos tab, you can view it with a sync'd/searchable transcript. It's a beta feature right now, and I know they'd appreciate feedback. Incidentally, while only a few of the oral histories have videos up, there's 500 or so transcripts up, will worth a perusal.