1995 article on the never-fully-developed Saturn 2

[parallaxscroll]

By whom? The product is called Sega CD. Unless someone has a time machine, this cannot be changed. Ever.

It was part of my revisionist history

The PC-Engine CD wasn't called NEC CD or Hudson CD was it ?

MegaCD was from MegaDrive, GenesisCD would be from Genesis..

[tomaitheous]

The GenesisCD gets extra 68000 CPUs, a much more powerful VDP and (either on VDP or seperate) scaling blitters, so its is on par with Sega's 'X Board' or 'Y Board' arcade hardware. This sounds expensive, but it wouldn't be for 1992. Less expensive than 32X of 1994.

That doesn't really make sense. If the 'GenesisCD' had these components, then there would no need for it to even attach to the Genesis in the first place. Attaching the Genesis wouldn't have anything to offer for it. The X and Y boards rivaled system32 and in some cases were even more flexible (source: Charles Macdonald).

I'd tone it back to something like: GenesisCD gets 16mhz 68k, 1024k of ram shared for 68k/ASIC, ASIC is setup for byte size pixels(256color) instead of nibble pixels(16color), a strictly frame buffer only VDP like in the 32x (the ASIC would render scaled sprites and such to the VDP's off screen buffer) - this is a low cost device and priority layering would be much more simple compared to the 32x version. You still need the Genesis' VDP for additional layers/planes and such. The memory on the GenesisCD should be interleaved so that the ASIC can grab two 16bit fetchs at once (32bit fetch for up to 4 pixels at once) thus no need for a speed increase and working with bytes instead of nibbles would reduce internal complexity for doing rotation 'fetches' and whatnot. Hell I'd also go as far as to reuse the EOL marker that the system16/23/X/Y boards used to speed up fill rate on sprites, on the ASIC. You loose one color, but you save some space and save cycles/increase fill rate.

[parallaxscroll]

That doesn't really make sense. If the 'GenesisCD' had these components, then there would no need for it to even attach to the Genesis in the first place. Attaching the Genesis wouldn't have anything to offer for it. The X and Y boards rivaled system32 and in some cases were even more flexible (source: Charles Macdonald).

I'd tone it back to something like: GenesisCD gets 16mhz 68k, 1024k of ram shared for 68k/ASIC, ASIC is setup for byte size pixels(256color) instead of nibble pixels(16color), a strictly frame buffer only VDP like in the 32x (the ASIC would render scaled sprites and such to the VDP's off screen buffer) - this is a low cost device and priority layering would be much more simple compared to the 32x version. You still need the Genesis' VDP for additional layers/planes and such. The memory on the GenesisCD should be interleaved so that the ASIC can grab two 16bit fetchs at once (32bit fetch for up to 4 pixels at once) thus no need for a speed increase and working with bytes instead of nibbles would reduce internal complexity for doing rotation 'fetches' and whatnot. Hell I'd also go as far as to reuse the EOL marker that the system16/23/X/Y boards used to speed up fill rate on sprites, on the ASIC. You loose one color, but you save some space and save cycles/increase fill rate.

I'd be okay with that,sounds fine. Some of that stuff is beyond my understanding, but I get the heart of what you're saying. Make the MegaCD/SegaCD/GenesisCD hardware something that's powerful, simple and cost effective. As long as the focus of the games is on arcade ports & newly created RPGs instead of FMV games, and everything it takes so it doesn't bomb, that's what matters most. What happened in the real history of things, Sega actually sold quite a few SegaCD units in the U.S.
Imagine if their overall strategy (hardware, software, marketing, price), had been better, it could've been a true success.

Then as long as the U.S. (and UK & Euro) Saturn contains the best possible CPU & 3D GPU hardware in one console (PowerPC 603e + Real3D/100) Sega would have a decent shot at 2nd place for that generation and could've sold maybe 20-30M systems worldwide and 10-15M in the U.S. (in real life Saturn sold 9.5M worldwide) helping to make sure Sega is on a good footing for the next generation.

[Joe Redifer]

It was part of my revisionist history

The PC-Engine CD wasn't called NEC CD or Hudson CD was it ?

MegaCD was from MegaDrive, GenesisCD would be from Genesis..

The actual PRODUCT NAME was Sega CD, so I see no logic trying to call it something else. It makes no sense. What other systems are called is irrelevant. Calling it the "GenesisCD" makes is confusing since it is not called that.

[parallaxscroll]

If you feel that strongly about it, we can just call it SegaCD (and MegaCD for Japan) as it was. I edited the name in my main post on the subject. What's more important of course, is the technology in SegaCD, the kind of games made for it, the price, and overall strategy for the expansion-upgrade to offer better, larger games on CD-ROM than what was possible on the base MD/Genesis.

[Joe Redifer]

Sega CD had a 12Mhz 68000, not 16Mhz. The power plants powering our homes at the time could not handle all of the power that a 16Mhz 68000 would require, and many black outs were caused when the Sega CD was released even with its wimpy 12Mhz CPU because even that required too much power. Way too much awesome for one household.

Anyway, the Sega CD could only rotate and scale at around 15-20 frames per second, maximum. Since these were all drawn as Genesis background tiles, there was no flicker in the scaled/rotated objects... ever.

[tomaitheous]

Joe: Stop trying to inject reality into this rewriting of history

Parallaxscroll: That's how I'd build the CD base unit. It didn't need 16mhz version, but I added it in for fun. And why not - 16mhz for 1992 wasn't a big deal for 68k's or other processors.

I have to admit though, SegaCD rolls off the tongue a bit better.

[parallaxscroll]

Found a website i remember reading years ago about 4 graphics processor choices from 3 providers that Sega had for Black Belt/Dural/Dreamcast

1, 2) Lockheed Martin Real3D/100, Real3D/Intel i740
3) 3DFX Banshee
4) NEC/PowerVR Second Generation

http://underworld.fortunecity.com/sonic/100/gp.htm

Among other things, they source EDGE's 1995 article on Lockheed Martin.


I'd say that both Real3D/100 (95-96 tech) and the much weaker Real3D/Intel i740 (97-98 tech) would've been too old, too underpowered for the Dreamcast of 98/99. The 3Dfx Banshee would be also. The PowerVR Second Generation (aka Series 2) and more specifically the custom PowerVR2DC / CLX2 variant, was the strongest chip in both rendering speed & rendering features. It was the best choice of the consumer 3D chips that would be ready in late 1998 when DC launched in Japan. That was the reality of the situation.

In my made up world of 'better everything', Sega & Lockheed would've made both a next-gen console and a MODEL4 arcade board. The next-gen console would push about 16-20M full-featured polys/sec and have pixel fillrate of 1 to 1.2 Gpixels/sec using a new 'Real3D/500' GPU. It would be about 30% or 1/3 as powerful as the new arcade board, but much more powerful than MODEL3. The MODEL4 would again be a multi-GPU board like MODEL3. It would be capable of about 50~60M polys/sec with 2x 'Real3D/5000' GPUs. The CPU would be a custom G4 with AltiVec running at 450 to 500 MHz. Main system memory would be 48 MB SDRAM on a 64-bit bus, and graphics memory would be 48 MB DDR on a wider 128-bit bus (the GPU needs the extra bandwidth). The audio would be pretty much the same chip that was in Dreamcast but with 8 MB DRAM. The optical disk could be either a 32x CD-ROM (cheap) or a 6x DVD-ROM (more expensive), but the main format for game media would be a much faster solid-state or magnetic disc/disc drive so that loading times can rival that of arcade machines. The downside would be smaller capacity (250 ~ 500 MB). There would also be a 10-12 GB hard disc drive. This machine is capable of doing most things a PC/MAC could do, including downloading music, video, web browsing, file sharing, basic productivity software. Japan is a key market for these non-gaming applications since most of their population do not own PCs.

Price: $1500, (one thousand five hundred dollars) LMAO