The History Of Nvidia GPUs (Archive)

[Deleted member 217926]

Hmmm, I've had a TNT 2 Pro, GeForce Ti 4600, 6600GT, 6800GS, 7600GT, 8800GTS 320mb, 9800GTX+, GTX 460 1GB, GTX 780 and now a GTX 980 Ti. All of them since the 6800GS have been EVGA. Pretty sure I'll buy an EVGA 1080 Ti when it comes out too.

I've also owned Ati cards starting with a 2mb Ati Rage II+ that came with my first PC in 1996.

 

[gondor]

You can use whatever construct you wish (polygons, meshes, etc) but they all get boiled down to triangles for final processing and rendering.

Having written rendering engines that use constructive solid geometry, splines, nurbs, etc., I believe your statement is nonsense. Even if you limit the scope of discussion to rasterizing systems, your statement remains false.

Well, kudos for writing 2D (?) graphics code but what he said is not nonsense, at least not in context written, and your comment, in spite of your chest thumping, contains very little substance to refute it.

The article mentions NVidia's dumb decision to use quadrilateral as the graphics primitive. Quadrilaterals consist of two triangles. Using triangle as the graphic primitive simplifies certain aspects of calculating things out, which in turn speeds up the processing. This isn't limited solely to rasterizing as there are many aspects to 3D graphics rendering prior to that.

 

[ravewulf]

I love these tech history articles.

For GPUs I've bought (as opposed to ones included with OEM machines), I've had the 8600 GTS, the GT 240, the GTX 650, and the GTX 750 TI. Mostly I just do video work and minimal gaming (newest I've got is Skyrim), so for the most part I try to stick to cards in the 60 - 75 watt range and get excited by the new video decode/encode support of the mid range chips.

 

[Bob_8_]

TNT and then the TNT2 - and then moved to the 3D Voodoo boards then added another 3D Voodoo board - I'm certain they were powered by my Intel Celeron 300 - OC'd to 450 -- with this huge house fan on the side of the cabinet,.,.lol., the fun days

 

[jsuryadinata]

Geforce 4, Radeon 9550, X300, 6600GT, 8800GT, HD5850, GTX960. From 1995 Duke Nukem 3D to DOOM 2016 computer graphic improvement mostly depend on graphic API. I remember how exciting when I get my HD5850 as it comes with directx 11 capability in 2011...however it takes sometime for the software to get mature and deliver something like DOOM 2016. Team RED and GREEN can always produce more powerful cards but if you want to see something different you'll have to wait for developer to maximize API capability properly.

 

[bloodroses]

I thought the NV1 was covered sufficient well. It was a pretty bold move, on Nvidia's part, to go for quadratic surfaces. However, I'd guess they failed to predict the impact of Direct 3D and its favor towards the lowest common denominator.

I also thought it was odd, because the graphics literature was full of bi-cubic patches and NURBS (non-uniform rational b-splines) surfaces. So, their decision to go with quadratics seemed doubly-odd. I assume it had to do with hardware implementation complexity and performance. Technically, I think it was a good choice. But, you need the game developers to come along with you...

However, I couldn't let this pass:

Needing a separate 2D video card was common in the 1990s

Can anyone tell me another card, besides the first two generations of 3dfx, that was 3D-only? This was not "common".

Also, I wish for separate board shots and screen caps. Or at least split-screen or picture-in-picture. The cross fades make it rather hard to appreciate either.

It is odd that Nvidia went for quads due to the standard leaning towards triangles. However, from what I've read about the Saturn system, the reason why quads was chosen was that more complex polygons could be created using less physical horsepower. The trade-off is that it is more complex on a software (mathematical) level to implement.

As with other 3D-Only cards, the only other one I knew of at the time was the PowerVR card. Maximum PC at the time (they might have been called Boot then) had built a computer that had a Voodoo, PowerVR and either a Nvidia or ATI card in it at the same time. This video shows another person who used a Voodoo 3, S3 Savage 4, and PowerVR PCX2:
https://www.youtube.com/watch?v=ADhvY2NW4sI

 

[bit_user]

It is odd that Nvidia went for quads due to the standard leaning towards triangles. However, from what I've read about the Saturn system, the reason why quads was chosen was that more complex polygons could be created using less physical horsepower. The trade-off is that it is more complex on a software (mathematical) level to implement.

Quadratic surfaces definitely enabled good looking models with far fewer patches than the equivalent number of polygons. The problem is that programmers and 3D modelling packages weren't used to dealing with them, and they required a lot of special-case code and game assets to be made for the NV-1.

As with other 3D-Only cards, the only other one I knew of at the time was the PowerVR card. Maximum PC at the time (they might have been called Boot then) had built a computer that had a Voodoo, PowerVR and either a Nvidia or ATI card in it at the same time.

Thanks for that. Does ring a bell. I still maintain than this wasn't normal. Most 3D cards & chipsets (S3, ATI, Matrox, Rendition, Nvidia, Number9, Cirrus Logic) integrated 2D and 3D into a single card/chipset.

I remember when boot magazine first launched. I think I actually had a subscription. I never knew they became MaximumPC, though. Another magazine I miss is Next Generation, but that was a bit more console-oriented.

 

[samopa]

Mine started with Monochrome/Composite Card, CGA Card. EGA Card, 8514/A Card, Tseng Labs XGA Card, 3dfx Voodoo Card, 3dfx Banshee Card (2d+3d version of Voodoo2), 3dfx Voodoo3 3000, 3dfx Voodoo5 5500, NVidia GeForce 4600 Ti, NVidia GeForce 6600, NVidia GeForce 6800 NVidia 8800GTX, NVidia GTX460, NVidia GTX480, NVidia GTX 580, NVidia GTX680, and now NVidia GTX980 Ti.
Boy, I should feeling really old now.

 

[bit_user]

■ Hercules (monochrome)
■ EGA
■ Tseng ET4000 SVGA
■ Cirrus Logic (some low-end PCI card, might've had 3D)
■ Rendition Verite (lost in the mail; never used)
■ ATI Rage 128
■ Radeon 9600 Pro
■ Radeon x1300 Pro
■ Radeon HD 4650
■ Radeon HD 6850
■ Radeon HD 7870
■ GTX 980 Ti

As you might guess, I've not been into PC gaming for a little while. The primary motivation for the last 3 cards was GPU compute and now VR.

Back when I did most of my 3D graphics programming, I had no hardware acceleration. Direct 3D didn't exist, and OpenGL was only available (on PCs) on Windows NT. So, I was limited strictly to mode 13h, for those in the know.

In fact, just to see 3D hardware in action, I once tried to arrange a visit at a SGI dealer. As I was quite young, the dealer quickly called my bluff and it wasn't to be.

 

[IInuyasha74]

As with other 3D-Only cards, the only other one I knew of at the time was the PowerVR card. Maximum PC at the time (they might have been called Boot then) had built a computer that had a Voodoo, PowerVR and either a Nvidia or ATI card in it at the same time.

Thanks for that. Does ring a bell. I still maintain than this wasn't normal. Most 3D cards & chipsets (S3, ATI, Matrox, Rendition, Nvidia, Number9, Cirrus Logic) integrated 2D and 3D into a single card/chipset.

I remember when boot magazine first launched. I think I actually had a subscription. I never knew they became MaximumPC, though. Another magazine I miss is Next Generation, but that was a bit more console-oriented.

Common was probably a slight exaggeration, but it was fairly common for most companies that developed graphics accelerators to create a 3D accelerator before producing a card capable of handling both 2D and 3D workloads. The Voodoo and Voodoo2 were by far the most famous 3D only cards. PowerVR's early cards are probably second. There was also the Matrox Impression product line and the relatively short lived S3 Trio3D, both of which were the first 3D graphics accelerators from each company.

Essentially everyone else that developed 3D graphics accelerators introduced products capable of processing both 2D and 3D images.

 

[DelroyMonjo]

Got 11 Nvidia cards dating back to the Riva 128. Skipped the 500 series. A pair of GTX 460's O/C'd were smokin'!....back in the day...

 

[aznchum]

The Geforce 7800 GTX 512 did not have a 512-bit bus, but rather a 256-bit bus.

 

[mitch074]

[...]
Common was probably a slight exaggeration, but it was fairly common for most companies that developed graphics accelerators to create a 3D accelerator before producing a card capable of handling both 2D and 3D workloads. The Voodoo and Voodoo2 were by far the most famous 3D only cards. PowerVR's early cards are probably second. There was also the Matrox Impression product line and the relatively short lived S3 Trio3D, both of which were the first 3D graphics accelerators from each company.

Essentially everyone else that developed 3D graphics accelerators introduced products capable of processing both 2D and 3D images.

No, the Trio3D wasn't a 3D-only chip; it was an S3 Trio64 with added (limited) 3D accelerated functions; it could accelerate single-texture-mapped triangles, but forget about shading and z-buffer. In pure 2D, it was a powerhouse, with excellent VESA 2.0/3.0 support, a fast RAMDAC and hardware-accelerated YUV-to-RGB conversion.

I first had a S3Trio64, supplemented with a 3Dfx Voodoo1; when I got my Celeron 300A, the S3 got kicked out and I put a TnT next to the Voodoo. When games started to support something other than Glide in 640x480, and the Cely couldn't cut it anymore, the Voodoo went away too - the TnT could still handle some stuff so I kept it a bit longer. I actually retired the TnT when GTA3 came out: it as a slideshow on that card. I replaced it with a Geforce4 Ti4200-8X, which itselft kept until I got a dual core (AMD Athlon64 X2 3800+) powering a passively cooled Geforce 6600. And that was the last Nvidia card I put in my main rig, as the RadeonHD 4850 came out and ran everything I threw at it for 4 years - and it burned out. After that, it was a RadeonHD 7770, then a R9-270X (I needed more oomph to deal with my WQHD screen) and just now a RX-480. Due to the shit my brother and friends had with their Geforces at the time, I became wary of Nvidia's stuff.

 

[Xoimbarg]

The Saturn used a crippled version. You could not use perspective correct texture mapping nor could you manipulate the control points on the the surface. They were basically just square quads. I am speaking as a developer for both. I had Descent working on the NV1 Diamond Edge cards. But making it work on the Saturn was impossible.

 

[Xoimbarg]

Another point, the over all architecture was cool, but everyone was writing and building 2d tori assets because we had to support non-accelerated computer systems. 3D accelerators were not that common so software render was king. Being able to do quadratic surfaces with 9 controller points was neat, but making a completely unique assets to show it off wasn't practical. Especially since all other accelerators and consoles were heading towards tri s

 

[nycalex]

i must be growing old, i actually recognize almost all cards on this list......

Riva TNT really was a breakthrough in PC gaming. i was using a matrox mystique (anyone remember this?) #d card + a generic 2d card. this used to be a nightmare to setup properly.
Riva TNT fixed all that by integrating 2d/3d.
used nvidia cards until ATI came out with the 9700 pro. All ATI/AMD until GTX 770 (gtx 680, same GPU).
i miss the glorious days of GPU competition, both companies were coming out with serious hardware.

but now......it's only nvidia. so i ended up purchasing a 1080. hoping that AMD catches up on the high end, they always give you more bang for your buck

 

[80-watt Hamster]

A polygon can be thought of as a simple 2D shape that, when assembling hundreds or thousands of these simple 2D shapes, forms a detailed 3D object. The most efficient polygon to use would be a triangle, because a triangle is made up of three points, and there is no way to arrange three points in 3D space in such a way that they exist on more than one plane. Three discrete points can only exist on one plane, whereas four or more can exist in multiple planes. Which is to say, any polygon more complex than a triangle can be expressed more simply as two or more triangles. Therefore, triangles became the standard polygon in 3D modelling.

Here's a very detailed and very good explanation on the subject: https://www.youtube.com/watch?v=KdyvizaygyY

Right, I get that. What's escaping me is why this paragraph:

"A unique aspect of the NV1's graphics accelerator is that it used quadratic surfaces as the most basic geometric primitive. This created difficulties for game designers to add support for the NV1 or to design games for it. This became increasingly problematic when Microsoft released its first revision of the DirectX gaming API, which was designed with polygons as the most basic geometric primitive."

... says that the NV1 was based on quads, while D3D was built around polys. Quads ARE polys. It's like saying your juicer doesn't work very well to make apple juice because it was designed for fruit. Sort of. I'm reaching a bit with that analogy.

 

[IInuyasha74]

The Geforce 7800 GTX 512 did not have a 512-bit bus, but rather a 256-bit bus.

Please check our review of the GeForce 7890 GTX 512.

http://www.tomshardware.com/reviews/nvidia,1171-2.html

 

[junkeymonkey]

one of the 3 founders

Jen-Hsun Huang (CEO as of 2013), a Taiwanese-born American, previously Director of CoreWare at LSI Logic and a microprocessor designer at Advanced Micro Devices (AMD).

so maybe some AMD tech involved ?? poor old AMD sems everyone capitalizes on them but them

 

[powernod]

Oh, the memories!!
I still have a Riva TNT2 Ultra stored in my shelf!! 😍

P.S. For the record:
Riva TNT2 Ultra, GeForce2 MX200, GeForce4 MX440, GeForce 6600GT, GeForce 7600GT, GeForce8600GT, GeForce 450GTS, GeForce 460GTX, GeForce 560Ti, GeForce750, GeForce 750Ti, GeForce 780

 

[bloodroses]

one of the 3 founders

Jen-Hsun Huang (CEO as of 2013), a Taiwanese-born American, previously Director of CoreWare at LSI Logic and a microprocessor designer at Advanced Micro Devices (AMD).

so maybe some AMD tech involved ?? poor old AMD sems everyone capitalizes on them but them

Jen-Hsun Huang helped found Nvidia in 1993 with 2 others. Chances are there wasn't AMD tech involved since at the time AMD was strictly a CPU company and Nvidia was strictly graphics. It was definitely a smart move Jen-Hsun's part as he saw a huge future in the graphics market; without Intel in that market to beat them down.

 

[Justiceinacan]

Haha, I still have an old PC with NForce 590 SLI chipset accompanied with 2x NVIDIA GeForce 8600 GTS. Memories of playing on that machine... it was quite nice for the time.

 

[dark_lord69]

My nVidia List:
1. Riva TNT
2. GeForce 256
3. GeForce 2 Ti
4. GeForce 4 Ti 4600
5. GeForce 6600 GT
"I stopped my nvidia "fandom" after geforce FX. Never again nvidia. No more going with the stupid flow."
Preach on brother!
--==As others have mentioned I was disappointed by the FX series so I moved to Radeon.==--
6. Radeon 1950XT
7. AMD HD 4870
8. AMD HD 7850
9. GeForce GTX 970

 

[dstarr3]

The Geforce 7800 GTX 512 did not have a 512-bit bus, but rather a 256-bit bus.

Please check our review of the GeForce 7890 GTX 512.

http://www.tomshardware.com/reviews/nvidia,1171-2.html

Wow. Way back when dual-slot cards were still kind of a niche thing, as opposed to the standard of today.

 

[bit_user]

Right, I get that. What's escaping me is why this paragraph:

"A unique aspect of the NV1's graphics accelerator is that it used quadratic surfaces as the most basic geometric primitive. This created difficulties for game designers to add support for the NV1 or to design games for it. This became increasingly problematic when Microsoft released its first revision of the DirectX gaming API, which was designed with polygons as the most basic geometric primitive."

... says that the NV1 was based on quads, while D3D was built around polys. Quads ARE polys. It's like saying your juicer doesn't work very well to make apple juice because it was designed for fruit. Sort of. I'm reaching a bit with that analogy.

First of all, note that it didn't just handle quadrilaterals, but quadratic (i.e. curved) patches:

https://en.wikipedia.org/wiki/Quadric

Of course, a planar quadrilateral is, in fact, a degenerate case of those.

Secondly, I found a good analysis of the NV-1, although the English is a bit rough:

http://vintage3d.org/nv1.php

omission of Z-buffer, which fired back upon NV1's d3d compatibility

This would be a big detriment, as many games probably required Z-buffering. Later, they point out two such examples (Carmageddon II and SOTE) that were relegated to D3D's software renderer.

Here's the more damning part:

Nvidia urged Microsoft to add support for quadrilaterals, but (now it is safe to say fortunately) they refused. As expected, this chip has extraordinarily hard time running d3d games. Triangles can be rendered when one side of quadrilateral has zero length, sacrificing quarter of vertex performance. The bigger problem is shading and texture mapping stages were not made for such special cases, often causing warping artifacts. Complete solution would also require pre-warped textures as it happened few times on Saturn, but PC developers were of course not going into such troubles to brighten Nvidia's day. Strangely enough, resolution affects the warping.

Direct3d driver came soon enough, wrapping triangles as quadratic surfaces for the chip. When d3d games actually started spreading, compatibility was finally exposed as minimal, missing important features like z-buffering, not to mention image quality issues.

The rest of that page is a good read (again, English aside). I actually had no idea quite how ambitious the NV-1 was. I'm guessing the Nintendo/SGI partnership on the N64, and the role of former SGI employees in early Nvidia might've been a big factor in the Nvidia/Sega partnership on Dreamcast. Elsewhere, I think I read the NV-1 debacle very nearly sunk Nvidia.

BTW, the Huang/LSI connection is also intriguing, as LSI was involved in the PS1. Not that I'm suggesting he knew this or used any of their IP for Nvidia - just pointing out the connection.