Should the SEGA DC and Saturn have launched with these alternative designs?

That doc reminds me just how beastly Real3D's design was back then. They decked that thing out to give it such sustained performance; that 750K tri/sec didn't come with much compromise.

PS2/PSP-era Sony graphics development continued on at their partner, Toshiba; the world wasn't deprived of its evolution. Sony simply found that an off-the-shelf nVidia solution was better than that to which CELL could be adapted for graphics and better than the RS that Toshiba specifcally designed to follow up their Graphics Synthesizer.

And still, yet another opportunity was given for Old Sony's vision of graphics evolution: Toshiba's team of GPU engineers turned around and brought their designs to the mobile market with the TC35711XBG.

http://www.toshiba.co.jp/about/press/2007_07/pr1701.htm

eye200708_01_1.gif


It's not considered a front runner for winning the PSP2 contract. When it's no longer able to expend so much extra power consumption, heat, or silicon than competing architectures, its true competitiveness in the open market becomes apparent.
 
Last edited by a moderator:
Interesting old press release ( not sure what it's got to do with Saturn or DC though ) - Are there any phones with it in?

I was thinking about the Saturn - the graphics chips in it are pretty much the same as the ( non Model 1/2 ) high end Sega Arcade boards, and it did a good job of reproducing the Model 1 / Model 2 graphics when coded by Sega's own teams. ( It was probably closer to PS1 performance than the DC was to PS2 )
 
I don't think spec-wise, it's disputable that the PS2 was a superior machine overall. Sony's libraries were notoriously bad initially so it wasn't readily apparent. By bad, I mean, it was a joke back then. Inafune at Capcom mentioned that "there aren't any".

Heh, I think that was done on purpose, to keep western developers in the backseat. Now with consoles basically using PC technology, western developers have the upper hand and are now dominating the marketplace.
 
I was thinking about the Saturn - the graphics chips in it are pretty much the same as the ( non Model 1/2 ) high end Sega Arcade boards, and it did a good job of reproducing the Model 1 / Model 2 graphics when coded by Sega's own teams. ( It was probably closer to PS1 performance than the DC was to PS2 )

Some of the model 2 conversions were awesome - Virtua Fighter, the Virtua Cop games, Last Bronx were all top stuff. After the first couple of graphically poor attempts like Virtua Fighter 1 and Daytona (not the later Championship Circuit Edition) they really got it together. As nice as Sega Rally and Daytona CCE looked though, the games that used VDP2 for lots of stuff aged better IMO.

I don't know what happened towards the end though, because Sega Touring Car was abysmal (later levels had unplayable low frame rates) and I remember House of the Dead looking bad. Maybe the later versions of Model 2 were just too fast, or maybe less resources were put into conversions.
 
So, anyway, the 32X. I'm trying to find out a bit more about it, but Google just keeps providing the same basic specs list (CPUs, RAM etc).

Anyone know anything about its VDP, or if it had the same bus contention issues for the CPUs?
 
So, anyway, the 32X. I'm trying to find out a bit more about it, but Google just keeps providing the same basic specs list (CPUs, RAM etc).

Anyone know anything about its VDP, or if it had the same bus contention issues for the CPUs?

CPU wise it was pretty similar ( master / slave SH2 ) - the graphics was pretty much a framebuffer, so not really that much contention.
 
Heh, I think that was done on purpose, to keep western developers in the backseat. Now with consoles basically using PC technology, western developers have the upper hand and are now dominating the marketplace.

This was certainly evident early on with games like MGS2, GT3, Ace Combat 4, Final Fantasy X, etc. Later in the system's life though I think Western developers were certainly pushing the PS2 harder than their Japanese counterparts with titles like Jak games, Ratchet and Clank series, God of War series, Killzone (I have a few choice words about dev decisions with this one!), etc. The follow ups to mentioned Japanese titles still improved upon their predecessors quite nicely though. MGS3 ditched the rectangular environment for a forested one. GT4 while marginal in overall graphics improvement, was 1080i compatible. Ace Combat 5 and Zero really improved the ground texture resolution, added tree models on the ground, as well cloud shadows on the ground. Final Fantasy XII broke out of the limited size environments into one that was much more open. While I don't think graphically the game looks too much better than X, it certainly was more challenging to implement such a scheme and they deserve credit as such.
 
SEGA actually tapped their occasional partner, Tantalus, for the Saturn conversion of The House of the Dead, an Australian outfit with the Saturn versions of Manx TT and the Wipeout games to their credit. They supposedly were rushed to release The House of the Dead with placeholder textures, accounting for the wretched, low resolution.

I'm not sure long-time partner Sharp nor Toshiba themselves have chosen Toshiba's TC35711XBG multimedia accelerator for their phones. With the "what if" discussions as to how future Sony systems would be different with a Ken Kutaragi-era Sony still around, the TC35711XBG's relevance to the topic is that it's the product of the evolution of those graphics initiatives.
 
Are you sure they were placeholder textures? I don't remember reading anything saying that.

One of the problems about Saturn 3D was that it ran via VDP1 - which had a fixed breakdown of it's VRAM into 2 256k frame buffers, and a 512k command/texture area. Commands took 32 bytes each - so if you have 3000 polygons in a frame you would need 96k to store the command list ( 192k if you double buffered it ) leaving only 416k or even 320k for textures. On the PS1 commands were stored in main memory ( sometimes in more compressed form ) so after taking 2 320x224 frame buffers you would still have 744k of memory for textures. So if you weren't using VDP2 for 2D backgrounds you would have around half the texture memory on Saturn.
 
That doc reminds me just how beastly Real3D's design was back then. They decked that thing out to give it such sustained performance; that 750K tri/sec didn't come with much compromise.

Real3D's designs were amazing for the time. Real3D/100 and Real3D-Pro/1000 were both 1995 hardware, intended for 1996 release. They wiped the floor with all competing designs in their classes.
 
I don't know what happened towards the end though, because Sega Touring Car was abysmal (later levels had unplayable low frame rates) and I remember House of the Dead looking bad. Maybe the later versions of Model 2 were just too fast, or maybe less resources were put into conversions.

Saturn was near the end of it's life cycle so Sega put less time and money behind the ports?
 
...

i just overclocked my dreamcast to 230 mhz and i am waiting for faster crystals to arrive :)
system is rock solid and performance is noticeably better, especially in Rez which showed a lot of slowdown in some parts :) ecco the dolphin is more fluid as well, especially in the larger levels :)
 
Seriously, this is the technology base Sega should've used for both Saturn (in 1996, not 1994/95) and Dreamcast (in 2000, not 1998/99).

mhsnxe.jpg


Real3D/100 is 1995 technology, with 3 main seperate processors (geometry, graphics, texture) pushing 750,000 fully featured polygons/sec. By 1996, Lockheed Martin Real3D could've combined the geometry processor, graphics processor and texture processor onto a SINGLE piece of silicon and run it at least 20% faster. I made up a name for such a chip, the "Real3D/120". It would've pushed 900,000 texture-mapped, perspective correct, gouraud shaded, lit, alpha-blended, anti-aliased, tri-linear filtered, z-buffered triangle polygons/sec. While it would not be as powerful as MODEL 3, it could've handled outstanding conversions with fewer polys. The most important thing is, because it would've been a single chip, it could've been manufactured and mass produced very cheaply. It would be a consumer chip then, not a workstation chipset/board. 3DO did similar with the M2, it was an expensive board full of many chips, then reduced into a single ASIC the Bulldog ASIC, with two PowerPC 602s at the front end. Sega could've used a 100 Mhz PowerPC 603, plus the sound hardware already developed for Saturn. It could've been done in 3 chips: PowerPC 603, Lockheed Martin Real3D/120, sound chip. This version of Saturn could've sold for $299 in 1996


For Dreamcast in 2000, Lockheed Martin Real3D could've designed a totally new generation of GPU, far beyond the Real3D/Pro-1000 used in MODEL 3, and far beyond the NV10 / GeForce 256, something that would blow away any in-house graphics solution Sony came up with for PS2, or at least rivaling the Evans & Sutherland RealIMAGE GPU that Sony might have concidered for PS2 if SEGA went with Lockheed Martin. The CPU would've been a modified PowerPC G3 (like Gekko in GameCube) but with a much more powerful vector unit, pushing around 3.2~3.8 GFLOPs. the Lockheed GPU would be 16 FPUs or 4 Vertex Shader like engines, providing the lions-share of the console's floating point performance. There would be 8 pixel pipelines with two texture units each. An 8:16 design. (PS2's GS was a 16:0 or 8:1 design). PowerVR could provide their ISP (image synthesis processor) for hidden surface removal, making the GPU much more bandwidth efficient and boosting the already high pixel fillrate (8 pipes X 180 Mhz = 1.4 Gpixels)) to massive effective rates. Also, TriTech (BitBoys) would also be involved in the Lockheed GPU, providing a bump-mapping solution. Overall it would be an 80-85% Lockheed Real3D design with the rest being contributions from PowerVR and TriTech.

This is all pure fantasy, dreaming and wishful thinking, but that's what this thread is about.
 
Last edited by a moderator:
Seriously, this is the technology base Sega should've used for both Saturn (in 1996, not 1994/95) and Dreamcast (in 2000, not 1998/99).

mhsnxe.jpg


Real3D/100 is 1995 technology, with 3 main seperate processors (geometry, graphics, texture) pushing 750,000 fully featured polygons/sec. By 1996, Lockheed Martin Real3D could've combined the geometry processor, graphics processor and texture processor onto a SINGLE piece of silicon and run it at least 20% faster. I made up a name for such a chip, the "Real3D/120". It would've pushed 900,000 texture-mapped, perspective correct, gouraud shaded, lit, alpha-blended, anti-aliased, tri-linear filtered, z-buffered triangle polygons/sec. While it would not be as powerful as MODEL 3, it could've handled outstanding conversions with fewer polys. The most important thing is, because it would've been a single chip, it could've been manufactured and mass produced very cheaply. It would be a consumer chip then, not a workstation chipset/board. 3DO did similar with the M2, it was an expensive board full of many chips, then reduced into a single ASIC the Bulldog ASIC, with two PowerPC 602s at the front end. Sega could've used a 100 Mhz PowerPC 603, plus the sound hardware already developed for Saturn. It could've been done in 3 chips: PowerPC 603, Lockheed Martin Real3D/120, sound chip. This version of Saturn could've sold for $299 in 1996


For Dreamcast in 2000, Lockheed Martin Real3D could've designed a totally new generation of GPU, far beyond the Real3D/Pro-1000 used in MODEL 3, and far beyond the NV10 / GeForce 256, something that would blow away any in-house graphics solution Sony came up with for PS2, or at least rivaling the Evans & Sutherland RealIMAGE GPU that Sony might have concidered for PS2 if SEGA went with Lockheed Martin. The CPU would've been a modified PowerPC G3 (like Gekko in GameCube) but with a much more powerful vector unit, pushing around 3.2~3.8 GFLOPs. the Lockheed GPU would be 16 FPUs or 4 Vertex Shader like engines, providing the lions-share of the console's floating point performance. There would be 8 pixel pipelines with two texture units each. An 8:16 design. (PS2's GS was a 16:0 or 8:1 design). PowerVR could provide their ISP (image synthesis processor) for hidden surface removal, making the GPU much more bandwidth efficient and boosting the already high pixel fillrate (8 pipes X 180 Mhz = 1.4 Gpixels)) to massive effective rates. Also, TriTech (BitBoys) would also be involved in the Lockheed GPU, providing a bump-mapping solution. Overall it would be an 80-85% Lockheed Real3D design with the rest being contributions from PowerVR and TriTech.

This is all pure fantasy, dreaming and wishful thinking, but that's what this thread is about.

It would have been lovely, but I doubt consumers would want to buy a new system for $1,000 at the time.
 
It would have been lovely, but I doubt consumers would want to buy a new system for $1,000 at the time.

That is what I was thinking while reading that post. It looks awesome and all, but no way could they have launched consoles with those specs and survived.
 
With massive mass production of single-chip GPU solutions, both Lockheed Martin Real3D + single PowerPC CPU consoles in 1996 and 2000, they could've both launched at $299, same price as PS1, PS2 and original Xbox.
 
One would imagine that if the R3D systems of the time were competitive they would have had design wins. Anyway, did R3D have any development capability then?
 
One would imagine that if the R3D systems of the time were competitive they would have had design wins. Anyway, did R3D have any development capability then?

Not that I recall. They were dinosaurs in 3D at the time, with their good graphics systems being highly expensive to make and costing boat loads of cash at the time. I really don't think they made too much off of R3D considering they weren't around too much longer after i740.

I do wonder if Megadrive's system could fit in one chip in 2000 and what the cost to make the thing would be. What kind of compromises would need to be made?

But in the reality of the situation R3D didn't have what it took to compete at the time. And that was having excellent bang for the buck while providing 3D graphics that nothing else could rival at the time for the price even computers priced much higher. Late 1998 was absolutely amazing to have that system launch. SEGA had many different partners to choose from, I feel they made the best decision at the time. It's just a no brainer, PowerVR was the only choice that made sense.
 
Back
Top