MS teaming up with ATI for the next xbox?

They were talking about the Dreamcast. Only thing that the NV1 and Saturn had in common were their rendering methods, and hence quick-to-port titles.

NV2
NVIDIA's financial savior came in the form of a video game console, more specifically, a Sega video game console. Through the NV1, NVIDIA had established a strong working relationship with Sega. The chip promoted sales of Saturn accessories and Sega programmers were somewhat familiar with quadratic surfaces after having ported a small number of games for the NV1. Most importantly, Direct3D was a non-issue because many of the Japanese console developers were ready and willing to use the unconventional technology of quadratic surfaces if it brought additional performance.
Sega funded a significant portion of the research on the NV2 and it is reasonable to suggest that NVIDIA might not exist it its capacity today if it were not for Sega's support of the NV2. Unfortunately, Sega eventually dropped the NV2 to give the Dreamcast a better future through an easier programming environment. Sega ended up going to 3dfx and later to PowerVR for the graphics technology in the Dreamcast. Little else is known about the NV2 story and the timing of events. Despite limited success of the NV1 and failure of the NV2, NVIDIA was not ready to quit

zurich
 
zurich said:
They were talking about the Dreamcast. Only thing that the NV1 and Saturn had in common were their rendering methods, and hence quick-to-port titles.

NV2
NVIDIA's financial savior came in the form of a video game console, more specifically, a Sega video game console. Through the NV1, NVIDIA had established a strong working relationship with Sega. The chip promoted sales of Saturn accessories and Sega programmers were somewhat familiar with quadratic surfaces after having ported a small number of games for the NV1. Most importantly, Direct3D was a non-issue because many of the Japanese console developers were ready and willing to use the unconventional technology of quadratic surfaces if it brought additional performance.
Sega funded a significant portion of the research on the NV2 and it is reasonable to suggest that NVIDIA might not exist it its capacity today if it were not for Sega's support of the NV2. Unfortunately, Sega eventually dropped the NV2 to give the Dreamcast a better future through an easier programming environment. Sega ended up going to 3dfx and later to PowerVR for the graphics technology in the Dreamcast. Little else is known about the NV2 story and the timing of events. Despite limited success of the NV1 and failure of the NV2, NVIDIA was not ready to quit

zurich

Oh, OK. Damn, now I really wish I could access FS from school... :(

Note to self, that means NV3 was Riva128, NV4 was TNT, NV5 was TNT2. OK, check.
 
NV4- TNT
NV5- TNT2
NV10 Geforce
NV11 Geforce2 MX
NV15 Geforce2 GTS
NV16 Geforce2 Ultra
NV17 Geforce4 MX
NV18 Geforce4 MX AGP8x
NV20 Geforce3
NV25 Geforce4
NV28 Geforce4 AGP8x
NV30 uh dunno :)

NV20_1 was Ti200
NV20_2 was Ti500 i think

Of course there was also TNT Vanta, TNT2 M64, Crush 18, Crush 11
 
ben6 said:
NV4- TNT
NV5- TNT2
NV10 Geforce
NV11 Geforce2 MX
NV15 Geforce2 GTS
NV16 Geforce2 Ultra
NV17 Geforce4 MX
NV18 Geforce4 MX AGP8x
NV20 Geforce3
NV25 Geforce4
NV28 Geforce4 AGP8x
NV30 uh dunno :)

NV20_1 was Ti200
NV20_2 was Ti500 i think

Of course there was also TNT Vanta, TNT2 M64, Crush 18, Crush 11

Knew all designations past GeForce256. Note the word 'past'... ;P
 
I've only read the first 2 pages of this thread, but a few points:

If MS is going to use an ATI chip for XBox2, it would NOT be in the R300/Radeon9700 generation. R300 would be ANCIENT by the time XBox2 is out (2005) More like an R500 at the very least. The main ATI design team who made R200/8500 is working on R400, and the ArtX team (team West I think) is working on the R500, AFAIK. If R400 is due out in 2003, and R500 in 2004, then R500 would be the minumim base for an ATI-based XBox2 GPU.

If MS chooses Nvidia again (the most likely of three main choices) then I would expect that at the very least, an NVidia-based XBox2 GPU would be in the NV4X family. Not the NV3X line. Anything that is NV3X based would be too old for XBox2. Depending on the pace of Nvidia developing new GPUs, I think NV40 would be the place to start XBox2 development, much like NV15/GeForce2 GTS (or NV16/GF2 Ultra) was the starting point for XBox development. Perhaps XBox2 will ultimately use an NV5X GPU, but I don't want to get too far ahead of things. If you remember, back in late 1999 "X-Box" was rumored to use an NV10 (GeForce256) and AMD K7 500 Mhz. - Then in early 2001, just before XBox was officially annouced, the GPU was said to be an NV15, but when it was actually annouced, MS said XBox would use an NV25, far ahead of the original predictions. Although ultimately, XBox ended up having the so called NV2A, which is inbetween
the NV20 and NV25, but still far ahead of the originally predicted (by magazines and websites) NV10.

So IMO, saying the XBox2 will have an NV3X GPU is probably like saying the XBox would have an NV10. The XBox ended up using something conciderably more advanced than that


The last possibility, is that MS might design its own GPU (and/or CPU)
I believe that this is the least likely possibility, but stranger things have happened. IIRC, MS owns the former CagEnt team. that is the design team that developed the 3D0/Matsushita M2 chipset and also the more powerful MX chipset (MX was basicly an M2 on sterioids) - this group of engineers got sold to Samsung around 1997 or so. then they almost got acquired by Nintendo, because Nintendo had their eye on the MX chipset, and wanted it to be used as the GPU of their next gen console, after a fallout with SGI over the N2000 project. CagEnt and the MX technology ended up in MS's WebTv division, if I'm not mistaken. So it seems MS does have some impressive graphics engineering talent and IPs. If they were to put that to use in XBox2, well that would be interesting.. the M2/MX people had done some amazing work that unfortunately never made it to market because of politics and Matsushita's cold feet.
 
Forgot to mention that ATI will probably work with Nintendo on the GPU for
GCN2 or (Gamecube2) this past spring, ATI mentioned at one of its conferences that its working on R400, R500 and was trying to win the contract for "Nintendo's next generation system" meaning beyond Gamecube.

- its news to me that Hitachi is working with Nintendo but I wouldn't totally surprize me. Perhaps Hitachi will supply the CPU for the next Nintendo (SH6 or SH7?) instead of IBM (as IBM-Sony-Toshiba are working on CELL which will be used in PS3 among other things) any links or anything more on Hitachi & Nintendo?
 
Oh, about Nvidia and Sega Saturn, there was NO Nvidia chip or Nvidia technology in Saturn at all. Although Nvidia and Sega did have a relationship. You see, Nvidia's first chip, the NV1, used in the Diamond Edge 3D card, had several Saturn games converted to it - VF Remix and Panzer Dragoon among others. The NV1-based Diamond Edge card was also made compatable with the Sega Saturn controllers. But Saturn was completely unrelated to NV1/Diamond Edge in hardware. though they shared some similar ideas on rendering, the develpment of NV1 was done by Nvidia and I think SGS Thompson as well. While Saturn was completely a Sega of Japan and Hitachi effort.

The Saturn had chips from Hitachi, and the VDPs (VDP1 and VDP2) were designed by Sega. What eventually became the Saturn actually started off (in 1991-1992) as the so-called Giga Drive, which was ment to be based on Sega's sprite pushing System32 arcade hardware, much like the Mega Drive (Genesis) had been based on the System16 hardware. Saturn went through MANY design changes throughout its development. The 32X is about what Saturn would have been, if it wasn't massively overhauled and upgraded again, when Sega learned about the PSX's performance. There was also no Lockheed-Martin chips in Saturn whatsoever. However, one faction within SoJ wanted to quickly scrap Saturn in favor of a Lockheed Martin design console, even before Saturn was launched in Japan on Nov 22, 1994. (I'll explain more about nV, LMC and Sega consoles later)
At the time that Saturn was in development, Sega only had arcade boards with General Electric and NEC chips in them (Model 1) there was no Lockheed Martin or Model 3 then anyway, in 1994.


Sega 3d polygon arcade hardware:
-Model 1 (1992) had General Electric Aerospace & Fujitsu 3D technology, NEC CPU
-Model 2 (1994) had Martin Marietta & Fujitsu 3D technology, NEC CPU
-Model 3 (1996) had Lockheed Martin Real3D technology, Motorola or Hitachi PowerPC 603e CPU
-NAOMI (1998 or 1999) Dreamcast based (PowerVR2DC + Hitachi SH4 with twice the video and main memory of DC.
-NAOMI2 (2000 or 2001) Dreamcast based with twin PowerVR2DC chips, ELAN T&L unit - 32MB memory each for CPU, ELAN and PowerVR2DCs, plus maybe 8 MB for sound - over 96 MB memory total

Real3D was a division of LM, it was the combined pool of technology and people from GE Aerospace's + Martin Marietta's graphics devisions, with Lockheed's as well. Saturn was designed in the years of 1992 through 1994, so all Sega had in 3D tech was the GE based Model 1 with flat shaded polygons. Model 2 was just coming into arcades as Saturn was being finished in 1994. The very most that Saturn could have possibily had, in the way of true 3D processors, would have been a stripped-down Martin Marietta chip or chipset used in Model 2 - but that would likely have been VERY expensive, thus Saturn didn't have it, or anything related to Model 2 at all. Saturn shared absolutely no technology with Model 1 or Model 2.

Martin Marietta bought GE's graphics division in 1993, then, Lockheed merged with MM in 1995 to form Lockheed-Martin Corp. Shortly after,
LMC put all of that graphics expertise from the combined companies under one roof, that was Real3D, a division or company within Lockheed-Martin - Real3D continued to work with Sega, in its new form, using its new Real3D Pro-1000 in Model 3 - and almost getting the contract to make a chipset (Real3D-100 or something else) for a Saturn upgrade, thus the so called VF3 upgrade cart, or a new system, or both. I read that both the upgrade and the new system where once called "Saturn2" by Next Generation magazine and some websites. (see NG Nov. 1995 for 2 page article)

Sometime before LMC and Sega were working to design a Saturn upgrade and/or sucessor, or perhaps during and after that time (its seems there was alot of back & forth/ on-again off-again with LMC/Sega) Nvidia was under contract to design a new graphics processor for Sega. The new graphics chip (it was in fact the NV2 that has been mentioned) was not for Saturn, but a Saturn sucessor, yet this was pre-Dreamcast. This's probably confusing the hell out of people, but.... anyway.... Sega's arcade groups liek AM2 didn't like what Nvidia was trying to push (quads), Sega wanted triangles but Nv wouldn't budge. Sega said, ok, that's fine, but quietly moved on to other (many other) chip makers. (LM, Matsushita M2, 3Dfx, PowerVR, etc)

Dreamcast was the direct result of Sega of Japan's PowerVR2-based console, first called Dural, then Katana, finally DreamCast, which competed with Sega of America's 3dfx-based "BlackBelt" console to become the next official Sega console beyond Saturn. Both the Dural/Katana/DC and BlackBelt prototypes came AFTER all co-operation with Nvidia had ceased. Probably LMC's as well. I like to compare this process of prototype console competition like this: you can think of Dural/Katana and BlackBelt as a face-off to be the console selected as Sega's sucessor to the Saturn, much like the USAF holding the ATF (advanced tactical fighter) competition between the Lockheed YF-22 and Northrop YF-23, to become the fighter that replaces the F-15 Eagle.
Its interesting that Lockheed was involved in one of the new fighter planes and an early design proposal for a new Sega console. The YF-22 won the ATF competition, and became the F-22 Raptor, kinda like the Dural/Katana won the console comparasion and became known as the Dreamcast :)

Ahem, anyway, the NV2 never got produced as far as anyone knows. at least not in any high-profile machine. The Firingsquad.com article is very very good, and seems to be accurate, (though the person that posted first about it in this thread got confused) that article said (or maybe it was another) that NV2 might have ended up in the Sega Pico, a childs learning toy that hooks up to a TV. whatever actually happened to the NV2, it was basicly a reject chip that Nvidia never talks about. It was quietly swept under the rug. However, as the article points out, Sega's funding of NV2 probably saved Nvidia from going down the tubes like so many other 3D chip makers. NV2 was not used in the Diamond Edge (again, DE 3D was NV1-based) Although NV1, NV2 and Saturn all used quads of some sort, but to be perfectly clear, neither NV1 nor NV2 were used in Saturn, or Dreamcast.

So these seperate technologies are:

NV1 - Nvidia/ SGS Thompson chip used in Diamond Edge 3D card
NV2 - abandoned chip or used in Pico(?) funded by Sega
NV3 - Riva 128 - Nvidia's first sucessful 3D chip - first DirectX Nv chip

Sega Saturn - Hitachi SH2 based with custom Sega video processors - vastly upgraded beyond original System32 base spec - somewhat more powerful than Model 1 in most areas - though not true polygon based like Model 1 or Model 2 - MUCH weaker than Model 2 - Saturn's 200k textured "polygons" not comparable to Model 2's 300k textured polys. Model 2 conversion to Saturn are no where near "arcade perfect" visually dispite what anyone says. best Saturn can do is maybe 30-40% graphically of Model 2. if that.

Lockheed Martin Real3D - used in Model 3 in the form of Real3D-Pro-1000
(2x) The fore-runners of LMC Real3D are the GE Aerospace & Martin Marietta chips, used in Model 1 and Model 2 respectively.

Dreamcast - Videologic PowerVR2, Hitachi SH-4 based. Not related to previous work that Nvidia did with NV2 for Sega, not related to any LMC tech either,
 
Then Intel bought Real3D and used that technology in their integrated chipsets for Pentium based motherboards.

That Starfighter thingy or whatever it was called ? Its the R3D-100, its the cheap version of R3D-1000PRO. If Sega had cancelled Saturn, and wait abit longer, it was rumoured that R3D-100 would power the Sega System.
 
That Starfighter thingy or whatever it was called ? Its the R3D-100, its the cheap version of R3D-1000PRO. If Sega had cancelled Saturn, and wait abit longer, it was rumoured that R3D-100 would power the Sega System.

Almost correct. the Starfighter card used the Real3D/Intel i740 chip. The i740 was also known as Auburn. It is not the same thing as the R3D-100 - the R3D-100 was a chipset with seperate processors: 1). a graphics processor (probably the main one for rasterizing) 2). a geometry processor (for T&L) 3). a texture processor (texture mapping)

The i740 was a single chip, and although it most likely shared some basic technology with R3D-100, it was not the same thing, i740 was less powerful overall. Where as the R3D-100 was used in low-end workstations or high-end PCs and was MORE powerful than the Martin-Marietta chipset in Sega's Model 2 board, the i740 chip in the Starfighter card was LESS powerful than Model 2 in realworld, especially framerates. Though the image quality of i740 was very good, its performance sucked, it was not as fast as Voodoo2, more like Voodoo1, but with better IQ than Voodoo2.


Many Sega enthusiasts such as myself thought that Sega would use use the LM R3D-100 in a Saturn upgrade, or even a new console. This was before the i740 was known about. Many people following the PC 3D world at the time thought that Lockheed's low cost entry into the consumer 3D graphics card market would be the R3D-100. that was not to be. it would take another 2 years for the i740 to appear in 1997 or so.
 
My ideal senareo for Sega's 32-bit generation would have gone something like this: Sega releases a 32-bit Giga Dive in 1993 in Japan, and 1994 in U.S. There is no 32X upgrade for Genersis. This Giga Drive is what it was originally suppose to be/should have been, in the real world - a simple combination of beefed up System32 sprite hardware and GE Model-1 polygon hardware. Giga Drive would be capable of around 200,000 flat shaded polys/sec, and these are real polygons, unlike the real Saturn.
Model-1-to-Giga Drive ports are painless and exact to the arcade. All System32 sprite games and Model-1 polygon games are brought over to Giga Drive during 1993-1995. Model-2 conversions might be attempted, but never released due to Giga Drive's inability to handle texture mapping or enough polygons.
When Sega learns of the PSX and Ultra64 specs in 1993-1994, they immediately turn to Lockheed-Martin to have them build a $200-$300 upgrade for the Giga Drive, with full 3D texture mapped polygons. This upgrade will fit on top of the Give Drive, like a turrent on top of a tank. The R3D-100 upgrade gives the Giga Drive at least 2x the performance of the Model 2 board (750,000 polys with texture and everything) more features than Model 2 (gouraud shading) better image quality, the ability to handle decent conversions of Model 3 games, even though the R3D-100 upgrade isn't as powerful as the twin R3D-Pro 1000 based-Model 3 (maybe 30-50%) though it can easily do better than exact Model 2 games. Even though this upgrade based on R3D-100 is more powerful than Model 2, it's MUCH cheaper to produce since it would be coming out in late 1996, (3 years after the base console Giga Drive is out in Japan, 2 years after it's out in the U.S.) where as Model 2 was built in 1993, being out in arcades in 1994's Daytona USA. Model-2 was so damn expensive because it was so old, 1993-1994, when 3D graphcis for any type of game, even arcades, was new, AND, because production was limited to several thousand units worldwide. Giga Drive's R3D-100 upgrade is produced in the millions, in late 1996 when better chip fabbing process is available and when ram prices are lower.

The R3D-100 handles most of the heavy lifting while Giga Drive's adaquate CPUs, audio chips, 4x CD-ROM drive, I/O, and power supply are still used.
A PowerPC CPU is not needed, btw, since Model 2 games didnt have one. The CPU(s) in Giga Drive are enough to handle Model 2 games when offloaded from 3D work done by R3D-100.
The combined GigaDrive + R3D-100 upgrade gives SEGA the most powerful machine of its generation. both in 2D AND in 3D. The 3D ability of this upgraded Giga Drive is many times more powerful than PSX, upgraded NEC PC-FX, Jaguar2, and several times more powerful than Nintendo Ultra64, somewhat more powerful than Matsushita's M2, and even beyond any PC 3D chip of the time from 3Dfx, Nvidia or PowerVR.. As well as having the highest image quality.

Sega's Giga Drive + LM R3D-100 upgrade is only matched by Lockheed's own R3D-100 cards for PCs, and surpassed only by Sega's Model 3 board. Then with the sucess of Giga Drive w/ its new found 3D texture mapping ability, its awesome upgrades of all Model 2 games, and decent conversions of some Model 3 games, and market place success, equaling PSX in sales, Sega hands the contract to Lockheed to design a R3D-500 for Sega's next gen console of 1998-1999, which will have 3D from the start! A new console with well beyond Model-3 ability, with around 10x the 3D power of the upgraded Giga Drive, in other words, about 3x Model-3 power, in the 5Mpps range. this catches Sony off guard, who has nothing to counter this new Sega console with until at least 2001. Sega goes on to win the 1998-2003 generation (being on par with Sony in the last) then partners with Lockheed yet again, and this time, MS, to build the XBox, which pulls together many of the major American graphics technology companies (LM, Nvidia, E&S, 3Dfx, Intel) to counter Japan's gaming industry, especially the 2001 released PS2, with XBox not getting released until 2003!

ahhh what never was... :) hey at least in reality we got PowerVR and they were decent!
 
Vince said:
Besides, Sony's *probobly* doing it with by combining the EE & GS on IBM's liecensed 0.10um SOI process and using it as an IOP or similar controller, I can hardly see MS conceiding parity and not matching this feature.

Wouldn't that mean MS would have to combine Intel's P3 and Nvidia's XGPU (and MCP?) all on one chip? That would be something of a first. Do you think Intel and Nvidia would allow that?
 
Right, Vince wrote that. Not sure how that got tangled up, maybe I was quoting Vince quoting you and accidentally forgot to delete all of your quote. I fixed it now.
 
Megadrive1988 said:
-Model 1 (1992) had General Electric Aerospace & Fujitsu 3D technology, NEC CPU
-Model 2 (1994) had Martin Marietta & Fujitsu 3D technology, NEC CPU

You are correct that Model 1 used an NEC CPU(V60 to be exact), but wrong about Model 2(2A-CRX, 2B-CRX, 2C-CRX).

Model 2 used an Intel I960+the same Fujitsu TGP chip as Model 1. 2A used the same but with 5 of the TGP chips. 2B used the I960+an Analog Devices SHARC DSP. 2C used I960+Fujitsu TGP4x (totally different from the other one).
 
Back
Top