Slashdot Mirror
Video Card History
John Mathers writes "While searching the net for information on old Voodoo Video Cards, I came across this fabulous Video Card History article up at FastSilicon.com. It's nice to see that someone has taken the time to look back on the Video Cards that revolutionized the personal computer. Here's a quote "When 3dfx released their first card in October of 1996, it hit the computer world like a right cross to the face. That card was called Voodoo. This new card held the door open for video game development. The Voodoo card was simply a 3D graphics accelerator, which was plugged into a regular 2D video card; known as piggy backing."
The Voodoo card was simply a 3D graphics accelerator, which was plugged into a regular 2D video card; known as piggy backing.
This isn't entirely correct, as any Voodoo 1 user could tell you. The card took up its own slot, and used a pass-through video cable to connect the monitor: When a Voodoo-compliant video signal was detected, it hijacked the output to the monitor and took over.
Nice design, for the time. The best thing was, it was CHEAP for the time (considering the performance). I think I paid $199.
M-
You catch enchiladas by picking them up behind the head and holding them underwater until they don't kick anymore -VeGas
I think what finally brought 3-D graphics acceleration into the mainstream was the introduction of graphics card chipsets that could combine decent 3-D acceleration with fast 2-D graphics all at once.
nVidia's pioneering RIVA 128 chipset was the first chipset that could compare itself in 3-D mode against the vaunted Voodoo cards of that period; once nVidia unveiled the groundbreaking TNT chipset it was pretty much over for Voodoo's separate board approach. This is what spurred ATI into developing the Rage Pro and later Rage 128 chipsets in the late 1990's, starting the competition between ATI and nVidia that has lasted to this day.
Erm. That's not even enough to fill in a single horizontal bar of the screen (unless you're running in 320*240 resolution). Perhaps they meant megapixels? This was hardly the only such error that I noticed, though - these guys really need to have someone proofread their articles.
--- Bwah?
What bullshit! The Rendition Verite supported bi & tri filtering, 32 bit color and a whole bunch of other 'now common' 3d features. The chip was well ahead of its time. It was the same problem nvidia first had. It had great features, but wasn't as fast as 3dfx. If Rendition would have released another card during the Riva128/TNT days (they did release the Vx2200.. which was nice, but a bit slow) with a tad more speed, we might be talking about Rendition, Nvidia, and ATi instead of just the latter two. All in all.. i can still remember playing VQuake, the first 3d version of quake, back when Carmack fully supported Verite and their far superior 3d technology.
Any article which try to encapsulate the history of 3d cards but fails to mention the Verite cards is a piss-poor article right from the get-go.
Check out Tom's Hardware Guide
http://www20.tomshardware.com/graphic/1997.html
http://www20.tomshardware.com/graphic/1998.html
http://www20.tomshardware.com/graphic/1999.html
http://www20.tomshardware.com/graphic/2000.html
http://www20.tomshardware.com/graphic/2001.html
When your CPUs floating-point throughput is a factor of 1000 better, that's when. In other words, at the rate at which general-purpose CPU technology advances, you'll be at that level of performance in about 15 years.
Stick Men
You beat me to it. It's so sad when people got excited about PC graphics cards. It wasn't/isn't because they were good, it's because they were /finally/ able to start doing what other platforms had been doing for years. Even then, the performance was poor - that's just when they started being able to display the same number of colours. The lowly Commodore 64 had better graphics than a PC with CGA graphics!
I note that the history of this article starts in 1996 . . . one year after Rendition's Verite chip became the first consumer add-on 3D accelerator
And absolutely no mention of Matrox whatsoever... despite the fact that their add-on 3D accelerator was arguably superior to the voodoo, and the parhelia is the ONLY 3d solution to support 3 display devices.
I am disrespectful to dirt! Can you see that I am serious?!
This card was massive and would never have been used in a server.
What about the early cards, TIGA, 8514/A & other 3D attempts like RIVA, Mystique, Virge? What about the cheats on PC benchmarks, back in VGA, now in 3D tests? What happened to Number 9, ELSA and other "Big" names in cards that are nolonger? Reads more like a Time magazine article then a serious attempt at a history of video cards Most glaring to me is the ATI 8500/Nvidia GF3 omission.
Which is a pretty simple way to get double the performance. I wonder why noone's done this recently
They are, in the chip itself, sorta. Modern all-in-one GPUs have multiple texture pipelines, which does split some of the load on the silicon level. It's not SLI, but it's the same concept.
The problem is SLI only doubles the fillrate. Both 3D chipsets need to work on the exact same dataset. SLI was a great boost back when 3D hardware was slow at texturing. These days the hardware can pump a couple thousand frames per second worth of textures, it's fancy multipass rendering and dynamic shaders (and to some extent, the geometry) that take up all of the frame generation time. SLI could speed some of this up, but it wouldn't help with most of the bottlenecks. It would be like putting new tires on a car that needs an engine tuneup.
Check this one:
http://www.accelenation.com/?ac.id.123.1
iD released a win32/OpenGL version of quake as unsupported, though free, software a couple of years after the original, called GLQuake.
EGA with 16 colours better than a Commodore Amiga? HAHAHAHA. In Ham mode, the Amiga was kicking out 4096! 16 colours are just garish. The Atari ST lead the charge, then the Commodore Amiga. The performance of the VGA graphics on my 386DX25 were dreadful. I added extra memory to my Paradise card so that it could handle 256 colours @ 640x480 under Windows and you had to watch it draw the screen line-by-line. The Commodore Amiga had been blowing it away for years by then. And for those who cared about improving the image on the Amiga, most of them went for a SCART connection rather than wasting their money on a monitor. PC owners didn't have a choice.
NVIDIA bought them out in December of 2000.
This text has some flaws... Nvidia didn't buy 3dfx nor its assets. It won them in a lawsuit with 3dfx.
-B
And I'll add only that Matrox basically invented (or at least first implemented in commercial product) video ram something like quarter century ago and that they had API capable of hardware accelerating 3d aps in a window in the times of win 3.11 (several years later Voodoo couldn't do it) :/
And no mention about that company whatsoever
But hey, what can you expect from (probably) fps kiddie biased negatively towards Matrox among others - because if he'd be JUST fps kiddie (not anti Matrox and...) he'd mention the fact that for ~half a year in 99 Matrox was the leader BOTH in performance and quality...too bad since then only the second holds true.
One that hath name thou can not otter
http://www.ati.com/support/driver.html
ATI generally releases an new WHQL Windows driver about once a month and a new Linux driver about every 6 weeks. I've had no problems with their XFree86 4.3 driver. They don't have a FreeBSD driver, though, but I guess a PowerBook would give somewhat of the same experience (BSD-based OS, XFree86-based X envrionment, Radeon 9600, plus Quartz/DisplayPDF and access to Mac apps). Mac OS X also has the ATI (and nVIDIA) drivers built-in and are updated with the software update utility.
ATI's Windows drivers are offically updated once in awhile, and are generally rock solid, but there are occasionally problems that aren't resolved for months at a time.
People say that the OS X GUI takes up too much CPU, while in fact it takes up almost none. All of the windows, shadows, etc, are being done with the video card though Quartz Extreme, no programming necessary from the app writer to take advantage of this either.
This isn't quite true. Most of the desktop rendering is still done by the CPU in the same way that it was done before QE was added to OS X. It's simply the individual windows that are rendered by QE, and OpenGL handles the 'surface' which is handed to it by Quartz and QE. So OpenGL mostly comes in to handle effects (like Expose, the fast user switching animation, and the opening/closing animations) and shadows, QE handles the windows and passes the textures to OpenGL when an effect is needed, and the CPU still does it's thing for the desktop (until you hand off the desktop as a surface to OpenGL for the user switching animation).
In other words, things that were basically eye-candy and were really slowing OS X down quite badly before QE are now handled by QE, but the base 2D engine that utilizes the CPU is still working the same way it does with most other operating systems.
That being said, free eye candy is free eye candy, and although there are many people out there that prefer things to be stripped down, I'd rather have something with a little flash that can be done just as quickly.
-PainKilleR-[CE]
i'm aware of two 3d accelerated versions of quake for windows back in the day: vquake, for rendition's verite line of cards, and glquake, which worked with any opengl compliant video card. glquake at first was used almost exclusively with 3dfx voodoo boards and so people thought it was 3dfx specific. in fact, 3dfx's voodoo card only supported a subset of the opengl api, hence they provided a "mini gl" driver that implemented only so much of the spec as glquake required. if you're fiddling with it today and don't have a voodoo card, you should remove the opengl32.dll in your quake directory - if memory serves, the one that came with glquake was 3dfx specific i believe and will load before your systemwide opengl library. both vquake and glquake were true win32 apps, so nothing of the dos based origins of quake should keep it from running in XP. and if it doesn't, i'm sure somebody out there has fixed it (the source is gpl'd now after all) and provides a patch and binaries for XP.
Note to self: pasty-skinned programmers ought not stand in the Mojave desert for multiple hours. -- John Carmack
It is possible to scale performance that way, but the result will be less than double the frame rate, simply because the time to generate a frame does not scale linearly with resolution.
To do 60 frames per second, you have roughly 16ms to generate a frame. A couple of those ms will be gobbled up with I/O transactions and various wait states, so you're already at the point where double the power is only going to result in 1.75x the performance. This will also be highly dependant on how well the 3D code can be parallelized (are there a lot of read_pixels callbacks that require both GPUs and both banks of memory to talk to each other? etc).
This has actually been done by SGI for awhile now. A couple years ago they took their Origin 3000 architecture and stuck on dozens of V12 GPUs and tiled the graphics for higher performance. That concept has been tweaked for their Onyx4 sytems... one large single computer with up to 34 ATI FireGL X1 GPUs. 16 GPUs work on each display in a 4x4 grid. Each GPU generates its 400x300 piece and 16 of those are composited in real time to make up a 1600x1200 display. I believe the biggest such machine to date has 32 GPUs powering two zippy fast 1600x1200 displays and 2 GPUs driving an additional 4 lesser powered displays. SGI gets quite a speedup by doing it that way, with 16 GPUs per display, but there's also a lot of overhead (even more in SGI's case, with 34 AGP 8X busses in said system). Their implementation of OpenGL and OpenGL Performer is tweaked for this, though.
So yeah, it can be done, but the fact that the GPUs will spend a significant amount of time doing non-rendering tasks (I/O, waiting for data, copying the final result to frame buffer, etc) means that you won't see a nice linear scaling. The cost of making custom hardware and custom drivers also adds up. With top-end PC 3D accelerators costing $400 already, I can't picture many users shelling out $1000+ for a dual GPU card.
Looks like DirectFBGL is 16 months old, although XDirectFB has a v1.0rc5 that's only 6 months old. I can't tell how you'd "make install" the two together, and whether existing apps would "just work", but someone else seems to be working on it.
--
make install -not war
Hi, Just got done instant messaging with the editor of Fastsilicon.com, Nathan Odle. He asked me to post here that he's pretty frustrated that the article was released without his editing, it wasn't ready yet and would have been quite different had his red pen gotten ahold of it. You can read the other articles on the site and see that this is the case - Nathan's standards for the content are VERY high, some heads are going to roll because of what happened here. I ask you ALL to check out the other content there, it's definitely well worth reading.
-JT
Actually there were 4 planes, and you could set a variety of resolutions besides 320x240 by tweaking the vga registers. One of my favorites was 512x384 because it was relatively high resolution with square pixels, but unfortunately it didn't work on all monitors. 320/360x400/480 were pretty stable and were used in Quake. 320x400 was used for the Win9x startup screen.
That must have been back around 1989? I remember seeing the first 386's with VGA graphics, and the demos featuring 320x200 256 colour graphics of a castle. By 1990, my paradise card was superceded by a $700 Hercules Graphics Station Card (TMS34010 processor) with came with 1 Megabyte of VRAM (the double-buffering option cost another $300) and four 24-bit colour images; A head-scan, a party-pup (don't even ask!!!), a fashion model and somebody leaning out of a window. I managed to write a SGI image format viewer, and thought viewing SGI's "helping build a better dinosaur" advert was the coolest image in my collection. At this time, Windows 3.1 didn't support 24-bit colour 3D graphics, so the only way to write your own extensions using TIGA.
,nVidia (1993), and Rendition (1993)
...
A rough sequence of the video standards would be:
1981 MDA,Hercules,CGA (IBM)
1983 PCjr (IBM)
1984 EGA (IBM)
1986 TIGA (Texas Instruments), 8514/A (IBM)
1987 MCGA,VGA (IBM)
1990 XGA, XGA-2 (IBM)
1990+ SVGA,XGA,SXGA,UXGA (various manufacturers)
1990+ VESA (manufacturers form consortium)
The early 1990's was probably the time of greatest change, when all the manufacturers were trying to outperform each other on resolution, refresh rate, video memory size, and finally 2D acceleration. The resulting chaos and incompatibility between cards led to the formation of the VESA consortium.
The mid 1990's were the time in which many of the innovators formed: 3Dfx (1994, from MediaVision)
NVidia introduced the NV-1 in May 1995,
3Dfx introduced the Voodoo card in October 1996.
ATI had been around since 1985, but didn't introduce the 3D Rage until September 1996 (but with No Z-buffer).
Matrox introduced the m3D in 1997 (piggybacking onto a VGA card).
Rendition introduces one of the first MiniGL drivers in 1998.
By 1999, each company was releasing new graphics cards every six months.
Any good gaming web site will give you the product history of each manufacturer.