in 1995-remember the announced $180 LockheedMartin R3D-100?

Back in mid 1995, there was an annoucment about Lockheed Martin bringing out a $180 (or $200) graphics card to the PC market, based on their new Real3D/100 chipset. It was later (years later) revealed that their PC/gaming chip was actually something else, the i740, a single chip that was co-developed with Intel. Though in 1995 (i740 was not known), many PC and 3D enthusiasts thought that the Real3D/100 chipset (consisting of three main processors: geometry processor, graphics processor, texture processor with performance of 750,000 textured polygons/triangles per second with healthy supply of features) would be Lockheed's entry into the highend-but mass-market PC card war in the same space as 3Dfx Voodoo Graphics and Videologic's PowerVR (and others)

The only two chipsets that really interested me back in 1996-1997, besides the PowerVR PCX1/PCX2, were Lockheed's Real3D/100 and TriTech's Pyramid3D (the version with the geometry processor)
I waited a long time for nothing, (i740 was ok, not great) and technology moved on. (Nvidia's TNT2/GeForces and ATI's Radeons now)

It was confusing in early 1997, a year and a half after the annoucment-- Lockheed had not been clear on what was going to be the $180 card. Later they said they had a single chip design for gamers. The i740 used in Starfigher cards and Intel. Though would it have been possible to manufacture the R3D/100 chipset for a $180 card? It turned out that REAL3D/100 would cost anywhere from $500 to $2000+ depending on how much video memory (upto 20-24MB)

The performance and image quality of this card was far beyond what 3Dfx Voodoo had in realworld performance (not so much on paper), naturally, since this was Lockheed Martin, who had all the combined talent, technology and experience from GE Aerospace's and Martin Marietta's graphics on military simulators. Far more than what an group that broke away from SGI had. It's a pity Lockheed did not enter the highend gaming graphics market. As well as TriTech's Pyramid3D.
 
Lockheed Martin ended up making Sega's Model 1, Model 2, and Model 3 hardware based on this technology, didn't they?

Edit: actually, maybe only Model 2 and 3?
 
To put simply, yes they did. (Lockheed merged with the companies that made Model 1 and Model 2, Lockheed was directly involved with Model 3) But all three arcade boards came before their consumer chip, the i740, and the professional chip many thought was going to be the new standard for desktop graphics, the Real3D/100.

Model 1, co-developed by Sega and General Electric Aerospace, first came out in 1992 in Virtua Racing, then VF.

Model 2, a souped up Model 1 with texture mapping, was developed with the assistance of Martin Marietta, first showing up in early 1994 in Daytona USA.

Model 3, a totally new design, was developed by Lockheed Martin's Real3D division, mostly finished by 1995, delayed a year, used first in Virtua Fighter 3 which released in 1996, but not seeing widespread use in games until 1997.

GE Aerospace (who was behind Model 1) was acquired by Martin Marietta (responsible for Model 2's texture-mapping) in 1993 i believe. Then Lockheed merged with Martin Marietta to become Lockheed Martin in 1995. shortly after that, Lockheed Martin created the Real3D graphics company/division to work on workstation/PC/arcade and possibly console graphics (console never happened) - Real3D was a combination of all the combined graphics tech and experience from the companies Lockheed got from GE Aerospace & Martin Marietta.
 
IIRC Model 1 was a NEC V60 and a collection of DSP's doing the 3d work.
The Real3d/100 was closer to Model 3 than Model 2. Model 2 only did 2 colours per texture!
Model 3 had all sorts of filters and blending modes and if games like Scud Race were anything to go by the IQ was certainly better than the original Voodoo.
 
I'm reaonsably certain that nearly all 3D graphics work at Lockheed Martin has ceased. The F-22's cockpit displays are going to use modified Quadro chips. While Lockheed is probably managing the redesign of the Quadro for military use, I doubt that they are going to be directly designing any 3D graphics chips (particularly now that they've been selected to design the Joint Strike Fighter...which should keep Lockheed going strong for a few more years...).
 
CMcK, I pretty much agree.

Model 1 was indeed an NEC V60 CPU (also used in sprite-based System32)
with alots of expensive (compared to consoles) DSPs doing all the 3D calculations. Model 2, IIRC, was a Model 1 with texture-mapping and more geometry pushing (300k pps textured compared to Model 1's 180k untextured) Model 2 did have bi-linear filtering I believe, that's why the graphics are not pixelated unlike Namco's System22 (Ridge Racer) which did not have texture filtering AFAIK. I do agree that Real3d/100 was closer to Model 3 tehn Model 2 (Model 3 had 1 or 2 Real3D-Pro/1000)
That is why I had hoped that the so called Saturn 3D upgrade (sometimes called Saturn2) had been a Real3D/100.

Even Model 2, tho it lacked gouraud shading and other features, was still superior to Voodoo1 in -some- aspects. the amount of realworld polygons for example. the geometry & textures that Model 2 pushed was just awesome. even with 2 color textures. Voodoo1 had more features, but was lower quality overall. regardless of paper technical specs, the Model 2 seemed more impressive than Voodoo1, but then Voodoo1 was 1995-1996 consumer tech, Model 2 was 1994 arcade tech. There is no question though, that the more powerful Model 3 wiped the floor with Voodoo1 and even Voodoo2.
 
Chalnoth: most likely, yes Lockheed seems to be out of designing more 3D graphics processors.

Question is, where did all of Lockheed Real3D's technology, patents and most importantly, its engineers, go? Intel got alot of it I believe, but I also heard that ATI got some(??), in addition to acquiring ArtX.

I recall a recent (within a year) interview with David Ortin of ATI, mentioning that he had been with GE Aerospace (and thus Martin Marietta and Lockheed I guess, since LMC merged with MM+GE)

Maybe Nvidia got some of Real3D's expertice and tech as well??
 
Let me just say that Lockheed still has TONS of need for engineers...perhaps many are still working with Lockheed, though not directly on building a 3D chip.
 
Chalnoth - Of course Lockheed needs tons of engineers for things like aircraft, weapons, rockets and other high-tech systems but I ment graphics engineers from the former Real3D division. Where are they now-
If Lockheed doesn't have them anymore?
 
Personally I think you are mistaken about the i740. I had one, it was a joke. I bet the guys at intel are still laughing their asses off about that one ;)
 
I had an i740 too.
It cost about Ă‚ÂŁ50 at the time.
It was out about the time of Voodoo2
I never had heard of OpenGL then even.. I thought it rocked.

Silly me!
 
megadrive0088 said:
Chalnoth - Of course Lockheed needs tons of engineers for things like aircraft, weapons, rockets and other high-tech systems but I ment graphics engineers from the former Real3D division. Where are they now-
If Lockheed doesn't have them anymore?

There's also a lot of computer power that goes into modern aircraft. Since Lockheed can't simpy use consumer-level chips (Only very recently has Congress decided that using consumer-level chips and modifying them was the best way).

To use all this computer power, Lockheed employs a large number of software engineers (I'm not privy to the specifics). Most of the chip design is outsourced, but you still need to have people experienced in chip design in order to properly program the hardware.
 
I'd hazard a guess that relatively wimpy processors go into most avionics.

Older MIPs and DSPs like i960 at the absolute top end. I think the shuttle orbiter still uses some VAX or PDP processors as flight computers. The ISS uses a lot of 386s.

As a data point, Intersil (used to be Harris Semiconductor) only has rad hardened 8086s available from their product listings.
 
RussSchultz said:
The ISS uses a lot of 386s.

Are you sure? I don't know about the ISS, but the Hubble telescope got a 486 upgrade ages ago.

Which of course does not contradict the main point, that space CPUs are old.

But, come to think of it - that is because you have to make sure they can stand the radiation, which is a long process. But there's not nearly as much radiation in an aeroplane as in a space shuttle. Shouldn't it be possible to get new processors inte aeroplanes much faster?
 
The exact chip used in Model 1 is the Fujitsu TGP MB86233. That is also the only thing preventing full emulation of Model 1 games-complete lack of info on the TGP chip. If anyone happens to know anything, feel free to contribute :)
 
I'm pretty sure about the 386's. I used to work for a division of Lockheed-Martin that did scientific payloads for Shuttle and ISS.

While our payloads had pentium class processors on them, the critical components of the ISS were 386's due to process (older process is generally more rad hard) and also because the on-chip cache was parity checked and the higher processors weren't (or a 386 doesn't have cache...I can't remember exactly).

As for military aircraft, search for "mil-spec processors" on google. It seems things are moving up and using more powerful processors.
 
Of course when you are talking about military equipment every single errata needs to be accounted for. You can't have an F22 seize up because your P4 decided to throttle its clockspeed.

Testing must take years in itself before the go ahead is given for a particular design to be used.
 
Well, I do know of one processor that the F-22 uses. It's an Intel processor designed for security (It has means of making memory very permission-based...so that no program can access any memory it is not qualified to access). It runs at 90MHz, if I remember correctly. No newer versions were ever made because Intel stopped production of the processor long ago. They expected the procesesor to take off for security uses, but it didn't, and so Lockheed is left with a rather old processor.

Together, all of the computer systems in the F-22 are pretty powerful. I was told about eight years ago that the two roughly tower-sized computing systems (Each had, if I remember, about 48 slots, each which had its own job, and I think each with at least one processor), that supposedly each had the processing power of a Cray supercomputer at the time.

I'm sure that they're not that powerful today by comparison, but I also doubt that a single modern PC could outmatch all the processing power placed into the F-22.
 
I wonder who has done the gpu of the recent i845g!
Must be the old 740-752 crew...
or who else?
 
I was kinda wondering the same thing. All or some of Intel's intergrated graphics (the "i" series) aren't they newer versions of the i740 which was based on Lockheed's Real3D technology?
 
Back
Top