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

[Secessionist]

DC:
CPU: SH5 @ 400 MHz (instead of the SH4 @ 200MHz) (faster CPU so it could do audio and T&L)
same GPU except at 135MHz instead of 100MHz.
2 MB cache buffer for the GD-ROM Drive; dual layer GD-ROMs (so the max size would be 2.4GB or 1/2 of a SL DVD).

Sega Saturn:
MIPS R4300i @ 100MHz (N64 CPU +6.25 MHz)

MIPS R4600 @ 100MHz (KI arcade board CPU)

SHARC DSP @ 100MHz (4 at unknown-to-me-clock speed used on Model 2 arcade board)

Pixel processor @ 50 MHz (does 32 bit RGBA blending @ 320x224/240 or 400x300 or 512x448)

quad speed CD-ROM drive with 1 MB cache buffer (instead of having double speed drive with half cache+SH1 controller)

8MB RDRAM (2x4MB on 16bit x 2 channel 500MHz bus) (same as N64 except double bandwidth to cancel out latency disadvantage)

I was wondering if the Saturn would've been more cost-effective that way with less processors that are more powerful.

 

[Nesh]

Whatever they would have chosen with the DC it wouldnt have helped.

They would have only increased costs and retail price.

As for the Saturn, considering how much costy it was for them to produce as it was, I am not sure if that would have had an affordable price while not eating away money from Sega's pockets.

A simpler acrhitecture enabling for a better software development environment instead of more performance might have helped much more.

The Saturn became what it was "at the last moment". There was no time and money available for experiementation. Tools werent ready and it was hard to get the performance out of it

 

[Secessionist]

You're probably right.

Sega should've worked on the software and developing tools more earlier on.

 

[eastmen]

There was nothing wrong with the dreamcast. by the time the dreamcast launched it was to late. The problems laid with the saturn.

The n64 cpu would have been to late, it launched what 18 months after the saturn . The model 2 cpu's would have been to expensive.

I think the 2 sh-2's with a model 1 gpu from power vr would have been their best bet , however I don't know if the model 1 would have been ready in time for the saturn.

The biggest mistake imo was the time line. If you look at history you see

Master system - Failure it sold 3m units against the nes
Genesis - Huge sucess , it cause the super ness to sell 20m fewer units than the nes and increased sega's sell through from 3m units up to the 20s I believe.

However then sega released the sega cd which was a failure , but then continued on with the 32x. Here in the states the 32x was just a year before the saturn and had few games released.

The Saturn - they surprised us all with a rushed launch with unfinished software. Look at vf that released with it compared to even vf remix . This coupled with the failure of the sega cd and 32x is what killed sega.

 

[brain_stew]

The hardware design was the least of the Dreamcast's problems, imo it was one of the most balanced and efficient console designs in history. SEGA didn't have the cash to spend any more money on the hardware anyway,a nd they sure as hell couldn't afford to increase the price by 50%.

 

[Nesh]

The "failure" of the Master System was due to Nintendo's agreements that binded developers to develop only for NES and the late government response against Nintendo's anticompetitive actions.

The Master System would have kicked Nintendo's butt if the developers were allowed to develop for it.

 

[function]

The Master System was a definite success in Europe, iirc outselling the NES up to the point the Megadrive came out. In the UK it was probably the most successful console between the 2600 and the Megadrive, and paved the way for all Sega's European success (that Sega later threw away).

The Megadrive was a big success, but SoJ torpedoed the still profitable machine to focus resources on getting a big Japanese win for the clearly doomed Saturn. SoJ were also responsible for the M-CD and made SoA poison Sega's image with the 32X.

The Saturn was an attempt to fit as many chips as possible inside a coffin. It was too late and too expensive to easily transition from the Megadrive, too weak to compete with the PS1 visually in all but a few carefully crafted first party titles, and it smelt of 32X and M-CD.

The Dreamcast smelt of Saturn, and Sega hired a crack management team to convince them to dump it (because physical platforms were about to die) just when they could have started making money from the online services the machine's expensive modems allowed them to provide.

The DC hardware was just about perfect though. More expensive, later-to-market hardware would have eliminated the slim chance to make it that they had by going first.

 

[kagemaru]

I remember back in the 32-bit days ("bits" HA!) where developers would say "this such and such effect/game shouldn't be possible on the saturn but we're pulling it off anyways". Such fond memories and the one factor that got me heavily interested in game development. For a system originally designed to pump out mostly 2D visuals, I was quite pleased with my collection of games.

Really wish Sega did design the Saturn better though. Could never see how they failed to see the industry heading to 3D when they themselves had the Model 1 board out in the arcades.

 

[Sonic]

I remember back in the 32-bit days ("bits" HA!) where developers would say "this such and such effect/game shouldn't be possible on the saturn but we're pulling it off anyways". Such fond memories and the one factor that got me heavily interested in game development. For a system originally designed to pump out mostly 2D visuals, I was quite pleased with my collection of games.

Really wish Sega did design the Saturn better though. Could never see how they failed to see the industry heading to 3D when they themselves had the Model 1 board out in the arcades.

In their defense, Model 1 was quite expensive in 1993. SEGA did think 3D was the future, just in arcades where they could sell expensive boards and cabinets for some nice profits, and back then their margins in arcades were really good. They did fail to foresee the home console market take to 3d graphics the way it did with the unveiling of the Playstation. We all know the last ditch efforts of putting in a second SH-2 core just to be able to compete. SEGA itself was best when it came to the Saturn and damn did NiGHTS and Panzer Dragoon Orta, Sega Rally etc. look really nice for the time.

The one thing I wish SEGA did back then is make Fighters Megamix high res like Virtua Fighter 2 was.

 

[Mobius1aic]

I wonder if releasing the DC in Japan and US at the same time could've helped. 2 years as opposed to just over 1 year before the PS2 release might have bought Sega more time to get an install base and compete. And I do agree that perhaps, if it was viable, that Sega should've pushed the clock speeds up. However, Sega didn't know what Sony was going to pull off, which of course, the PS2's announcement stole thunder from the DC's launch in the US in '99. Except in a couple areas, the PS2 was just way ahead of what the DC could pull off. I still wonder if the DC should even be considered part of that same generation of consoles. It was more like an intermediary, much like the 3DO and Atari Jaguar were to the PS1. Though it's funny, the DC to the PS2, is not all the different from comparing the PS2 to the Xbox. The Xbox was released over a year later and vastly more powerful (and a huge cost to MS). PS2 had the Sony name and the software to sell it though. The GC was of course the intermediary between the PS2 and Xbox, but at least benefited from some nice balancing like the DC did, was cost effective to produce. It's only too bad that most developers treated it like some underpowered system best for children (which we can blame Nintendo much for).

 

[eastmen]

I wonder if releasing the DC in Japan and US at the same time could've helped. 2 years as opposed to just over 1 year before the PS2 release might have bought Sega more time to get an install base and compete. And I do agree that perhaps, if it was viable, that Sega should've pushed the clock speeds up. However, Sega didn't know what Sony was going to pull off, which of course, the PS2's announcement stole thunder from the DC's launch in the US in '99. Except in a couple areas, the PS2 was just way ahead of what the DC could pull off. I still wonder if the DC should even be considered part of that same generation of consoles. It was more like an intermediary, much like the 3DO and Atari Jaguar were to the PS1. Though it's funny, the DC to the PS2, is not all the different from comparing the PS2 to the Xbox. The Xbox was released over a year later and vastly more powerful (and a huge cost to MS). PS2 had the Sony name and the software to sell it though. The GC was of course the intermediary between the PS2 and Xbox, but at least benefited from some nice balancing like the DC did, was cost effective to produce. It's only too bad that most developers treated it like some underpowered system best for children (which we can blame Nintendo much for).

The ps2 wasn't way ahead of the dreamcast. The false promises and cgi is what killed the dreamcast and the fact that sega was bankrupt

 

[Mobius1aic]

The ps2 wasn't way ahead of the dreamcast. The false promises and cgi is what killed the dreamcast and the fact that sega was bankrupt

Except in the texturing department, what I see pushed out of the PS2, even early on was way ahead of what I ever saw pushed out of the DC.

 

[SG79]

The ps2 wasn't way ahead of the dreamcast. The false promises and cgi is what killed the dreamcast and the fact that sega was bankrupt

As someone who got the DC in late 98, I can tell you that Sega a lot of boneheaded decisions from day one in Japan. Regardless, that's off topic so I'll stop there :neutral:

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".

I think Namco's own SC and TTT demonstrated the differences well. SC had superior animations (being based on a newer tech from 98 vs. TTT's T3-based animation from late 96), and subtle, well placed lighting. TTT had the better texture fidelity, much higher poly count and more complex lighting, though the lighting did look exaggeration and cartoony.

The geometry and poly count in games like MGS2 were way ahead of anything on the DC back then.

 

[Nesh]

The ps2 wasn't way ahead of the dreamcast. The false promises and cgi is what killed the dreamcast and the fact that sega was bankrupt

Define the quality of the CGI. For many of us the real time tech demos that built the hype looked CGI and unreal for many including me at the time of first reveal. Thats the quality we were expecting. Not CGI at the level of Spirits Within or FFX CGI quality. The most recent CGI quality is always an unreachable target.

To name a few examples, GT3 and Tekken Tag which came early in the console's life cycle were real games that proved that the original GT and Tekken tech demos that developed some of the hype against the DC were doable. These were only some of the games that reached our expectations after the hype in terms of graphics

These two games alone pulled me away from the DC. And these are just two examples

And MGS2 simply made my jaw drop when they showed the first trailer. It was as if everything Sony demonstrated with the tech demos came to fruition.

 

[SG79]

Speaking of the Tekken demo:

http://www.youtube.com/watch?v=eJsi-SgksHY

It was playable, and the models look about the same as TTT.

 

[Mobius1aic]

GT3, The Bouncer (as mediocre as it was), FFX, Ace Combat 4, and of course......MGS2 all pretty much shut the lid down on the DC in terms of possible graphical parity with the PS2, especially in the geometry department. Even to this day, I still think The Bouncer had some of the best geometry in any PS2 game. Sure it kind of sucked, but it sure was pretty. For fun though I would love to see someone do a real comparison of how the PS2 and DC handled texturing versus each other considering the DC had dedicated texture memory and if I'm not mistaken, hardware texture compression.

Funny, supposedly according to this, each texture layer on the DC had to be applied as another polygon, though the writer says if programmed right, it was done efficiently. Also I'm glad to see another piece of proof of the DC's bumpmapping and shadow volume capabilities.

 

[function]

Software always lags behind hardware, and it's a great pity that the DC's graphics chip only seemed to be used for poly + texture, even by Sega in their megabudget Shenmue. It would be unthinkable now to leave off bump mapping (or better) on a big budget graphical showcase, but I guess it was unthinkable back then to use it. If you're building your engine, tools and assets before you actually get a finished GPU it probably doesn't help.

The DC was going up against the newer, more expensive, and silicon packed PS2 with some of its biggest guns locked away. Simon F said performance couldn't have been the reason the DCs bump mapping never got used, and multi-million pps bump mapped demos would seem to validate this. It's disappointing to see claims that the Dreamcast was "maxed out" by launch titles and that there would have been nowhere for improvements if the machine had lasted a normal length of time.

And I do agree that perhaps, if it was viable, that Sega should've pushed the clock speeds up.

Yeah, delaying or massively increasing the complexity of the system with multiple CPUs was a definite no, but accepting increased power consumption in order to run the system (in particular the CPU) faster would have made sense.

Modders have had some success at overclocking the CPU - this one is up to 270 mHz with only a small heatpipe leading to a tiny radiator while still in the original case (perhaps it's using the launch units cooling system?):

http://www.youtube.com/watch?v=CtVK-vZZG1M

A bigger case with a small aluminium heatsink and 4cm fan would make more sense if you were doing it on a final product. A 250 - 300 mhz CPU could have offered a big jump in performance for only a small increase in cost. But like you say, Sega didn't know Sony were going to land with such a silicon heavy and (relatively) power hungry competitor.

If you really wanted to add more silicon to the DC then it might have been best to play to the system's strengths and increase the video ram and have games awash with high res textures and bump mapping (making sure to develop tools to help people use it). Try and overload scenes with detail that way - you'd never be able to match the PS2's poly counts and vertex lighting ability, but so much of what makes a game look good or look detailed is related to texturing and texturing effects that you'd still stand a good chance of being able to make stuff that looked equally impressive.

 

[Simon F]

Gconsidering the DC had dedicated texture memory and if I'm not mistaken, hardware texture compression.

DC's CLX did not have "dedicated texture memory". It shared 8MB (?) for frame buffer (and back buffer), textures, and geometry. It did have a texture cache, if that's what you mean, but that was all transparent to the developer.

It certainly had texture compression which was (asymptotically) 2bpp (or even 1bpp -but the texture needed to be quite simple for that level of compression). More impressive, in my opinion, was that CLX had full hardware translucency sorting. When you look at the back-flips that today's programmable hardware has to do to achieve the same it can leave you feeling sad.

Funny, supposedly according to this, each texture layer on the DC had to be applied as another polygon, though the writer says if programmed right, it was done efficiently. Also I'm glad to see another piece of proof of the DC's bumpmapping and shadow volume capabilities.

IIRC, with the possible exception of one or two 3Dfx systems which possibly could support two texture layers (provided Trilinear wasn't being used) I don't think any of the systems of that era directly supported multiple texture layers per polygon.


Incidentally, there is some incorrect information in that link you gave. I'm pretty certain that the shadow/light/modifier volumes do not have to be convex but they probably have to be closed and the fill rule is probably odd-even.

 

[Squeak]

Except in the texturing department, what I see pushed out of the PS2, even early on was way ahead of what I ever saw pushed out of the DC.

It's still a mystery to me why the PS2 wasn't better at texturing. It seemed to have all things going for it.
- It did have texture compression, (palettized modes) even to the same ratios of the DC, though in some ways not as refined.
- It did have a large piece of the VRAM for textures (512 - 1512Kb depending on the buffersize), that with even semidecent, coarse virtualization of the textures should have been enough to do gobsmacking textures both in resolution and variety.
- And it did have the very capable IPU MPEG2 decoder, that could have been used for high quality 16 or 32bit textures (2-4x essentially lossless), that could have had appropriate pallettes connected to them after decompression to extract two or four 8bit textures or even more 4bit textures.

It did have trouble with inclination based MIP mapping, and the loss of efficiency because of that (texturing thousands of texels that aren't visible, because of the slant of the surface and resulting in shimmering), and according to some developers, that was the main problem. I find that pretty hard to believe though. And if it was such a big problem why wasn't it worth fixing in software, even if the cost would be quite substansial?

 

[Squeak]

In their defense, Model 1 was quite expensive in 1993. SEGA did think 3D was the future, just in arcades where they could sell expensive boards and cabinets for some nice profits, and back then their margins in arcades were really good. They did fail to foresee the home console market take to 3d graphics the way it did with the unveiling of the Playstation. We all know the last ditch efforts of putting in a second SH-2 core just to be able to compete. SEGA itself was best when it came to the Saturn and damn did NiGHTS and Panzer Dragoon Orta, Sega Rally etc. look really nice for the time.

The one thing I wish SEGA did back then is make Fighters Megamix high res like Virtua Fighter 2 was.

I think acceptable 3D for consumers could have been done much earlier, like 85 or 87. Maybe not with that much texturing, but still more than good enough to do an I, Robot, Hard Drivin' 3D or better universe and have it run at 30 or 60 fps. All it needed was some dedicated hardware designed for the job.
Considering how well the aforementioned arcade games did with early DSPs and bitslice tech I think a dedicated transformation processor (optimised multiply accumulator) connected to a decent CPU (68000 or 65816) could have kickstarted non-novelty 3D years before it happened with the Playstation.