I don't think you're wrong at all about the PPC603+R3D/100 setup. Unless I misunderstood you.
That's exactly the setup Next Generation mentioned in their article.
As for M2 (fantastic articles BTW, thx!) I know the BDA had a triangle engine. I'm not sure if that could produce its own geometry without relying on the PPC602 CPU(s). Would be interesting if it could. Again, I'm not knowledgeable enough to know.
I am going to speculate that M2's successor, the MX, had a geometry engine, but I just don't know. It was described as 'M2 on steroids'. There are only a few details on MX. One interesting thing about MX, at least one version of it, is that it was going to use embedded memory on the graphics chip. This was a couple of years before PS2's Graphics Synthesizer and Gamecube's Flipper, both of which had embedded memory on their graphics chip/GPU respectively (GS wasn't a GPU).
https://groups.google.com/forum/?fromgr ... 7itfq2pbYJAs for MX (see IG's Fusion issue 10), the current concept being tossed around is the idea of actually including the video RAM frame buffer within the actual MX chipset rather than externally -- as transferring data from separate RAM chips to the math processors is one of the most vital time delays in any computer or game console, having the RAM bundled with the fast MX chipset would mean incredible speedups in processing. Developers claim that such an MX chipset could deliver -- believe it or not -- 15-20 million polygon per second performance.
The drawback? The failure rate of such combined chips could be prohibitively high -- between the RAM and the high-intensity math processor, the chips could fail in production at a rate of 20% or greater depending on how much RAM was included on a chip. Additionally, the heat generated by such a configuration would mandate special cooling measures. Regardless, the premise is food for thought and some additional RAM may well wind up in the final MX design.
https://groups.google.com/forum/?fromgr ... Mu7flmi7cJAlso mentioned in the article is how two new chipsets are supposedly under development. One is an enhanced M2 codenamed MX and is described as 'M2 on steroids'. BTW it is mentioned that the M is really most likely a common moniker used for version 2 type projects meaning it stands for mark, thus 'Mark 2'. And in MX the X is obviously a variable. MX so far 'offers twice the performance of the M2 chipset...currently intended for PC and arcade use...'. Finally the totally newer chipset is codenamed S42 - S being just another letter like M and 42 being the one calculated as the meaning of life by the computer Deep Thought in the Hithchiker's Guide to the Galaxy. S42 being post 64bit era forecasting that probably is the equivalent of the M2 when the Opera was being made.
If we are to take this seriously, then MX = M3 and "S42" = M4, with S42/M4 existing only on paper.
Interestingly enough, Nintendo wanted to use the MX as the basis for their next-gen console (which the press called N2000) after their falling out with SGI, before hooking up with ArtX. Except that instead of a PowerPC CPU, Nintendo would use MIPS with the MX graphics tech. Ironically, Nintendo went to IBM for the Dolphin's CPU. I hope that's not too confusing.
Here's a great article everyone should read.
https://groups.google.com/forum/?fromgr ... Dd2vV6R7QJIs Nintendo in Trouble?
Although experts acknowledge that the video games business is surprisingly
incestuous by even Jerry Springer’s standards, recent developments taking place within two of Seattle’s biggest corporations have made that fact clear for the whole world to see. Next Generation Online exclusively reports on how Nintendo and Microsoft wound up eyeing the same company’s chipset for the year 2000’s biggest game console.
Few in the video game industry are aware of a rift that formed between Nintendo and partner Silicon Graphics, Inc. just as their jointly-developed 64-bit game console rolled off production lines. Already beginning to feel financial strains due to changing market conditions for their high-end graphics workstations, Silicon Graphics found itself arguing over component profits with notoriously tight-fisted Nintendo as the system’s American launch MSRP was lowered at the last minute before release. Although the companies maintained their working relationship, the decidedly traditional and hard-lined management at Nintendo had taken offense, and no longer considered SGI a lock for development of Nintendo’s post-N64 game console.
Then several important events took place during 1997 inside of Nintendo, SGI and one of their former competitors. Weak Japanese sales of the N64 and its software lowered the company’s confidence in the N64 platform, and American sales were projected to fall off as key internal software titles were continuing to miss release targets by entire seasons. Demonstrably strong sales of PlayStation games in the inexpensive CD format had weakened the appeal of Nintendo’s third-party development contracts, and Nintendo started to believe that it was in the company’s immediate interest to prepare a new console for release as soon as Fall of 1999. At the same time, a number of Silicon Graphics key Nintendo 64 engineers left the company to form the new firm ArtX, with the express intention to win a development contract for Nintendo’s next hardware by offering Nintendo the same talent pool sans SGI’s manufacturing and management teams.
As it turns out, most of the industry’s top 3D chip experts have been lured away from smaller firms by accelerator developers NVidia, 3Dfx and NEC, so Nintendo’s pool of potential partners was already shrinking when it began to shop around for a new console design team. Enter CagEnt, a division of consumer electronics manufacturer Samsung, and here’s where the confusion begins: CagEnt was formerly owned by 3DO, where it operated under the name 3DO Systems and developed the M2 technology that was sold to Panasonic for $100 Million some time ago. When 3DO decided to exit the hardware business, it sold off the 3DO Systems division to Samsung, which named it CagEnt and gave it roughly two years to turn a profit. CagEnt owned three key technologies: a DVD playback system, a realtime MPEG encoding system called MPEG Xpress, and a completed game console with a brand new set of console-ready chip designs called the MX. Adrian Sfarti, who had formerly developed the graphics architecture design for SGI’s Indy workstation, was the head of the MX project.
The MX chipset was a dramatically enhanced version of the M2 chipset sold to Panasonic and Matsushita, now capable of a 100 million pixel per second fillrate and utilizing two PowerPC 602 chips at its core. (CagEnt’s executives also boasted of a four million triangle per second peak draw rate, though the quality of those tiny triangles would of course have been limited). Nintendo executives Howard Lincoln and Genyo Takeda were among a group of visiting dignitaries to tour CagEnt’s facilities, culminating in late 1997 or early 1998 with a formal offer from Nintendo to acquire CagEnt outright. At this point, Nintendo had terminated its development contract with SGI (see SGI/MIPS Loses Nintendo Business).
As purchase negotiations continued, Nintendo worked with CagEnt engineers on preliminary plans to redesign the MX architecture around a MIPS CPU, as Nintendo’s manufacturing partner NEC has a MIPS development license but none to produce the PowerPC 602. Nintendo and CagEnt flip-flopped on whether the finished machine would include a built-in CD-ROM or DVD-ROM as its primary storage medium, with Nintendo apparently continuing to insist that ROM cartridges would remain at the core of its new game system. Yet as DVD and MPEG technologies would have been part of the CagEnt acquisition, Nintendo would probably have found some reasonable use for those patents eventually. The MX-based machine was to be ready for sale in Japan in fall 1999 -- in other words, development of games for the new console would begin within literally months, starting with the shipment of dev kits to key teams at Rare and Nintendo’s Japanese headquarters.
Although the asking price for CagEnt was extremely low by industry standards, talks unexpectedly broke off in early 1998 when Samsung and Nintendo apparently disagreed on final terms of CagEnt’s ownership, leaving Samsung’s management desperate for a suitor to buy the company. CagEnt aggressively shopped itself around to other major industry players. SGI’s MIPS division, reeling from the loss of its N64 engineers to ArtX, allegedly considered acquiring CagEnt as a means to offer Nintendo the technology it had already decided it liked. Sega, 3Dfx and other companies toured CagEnt’s facilities and finally CagEnt found a suitor.
In early April, Microsoft’s WebTV division ultimately acquired all of the assets of CagEnt and hired on most of its key personnel. WebTV and Microsoft apparently intend to use the MX technology at the core of their next WebTV device, which as might be guessed from the graphics technology, will no longer be limited to simple web browsing and E-mailing functionality. The next generation WebTV box will be Microsoft’s low-cost entry into the world of game consoles, melding the functionality of a low-end computer with a television set-top box and game-playing abilities. Having worked with Sega behind the scenes since 1993 or 1994, Microsoft has been quietly gathering the knowledge it needs to market and develop games for such a device, and now it has the hardware that even Nintendo would once have wanted for itself.
As for Nintendo, all signs point to a very unpleasant near future for the Japanese giant. Lacking internal hardware engineers with the necessary expertise to develop the next high-end chipset, Nintendo is now all but forced to either partner with ArtX, or one of the 3D accelerator makers who have been sucking the industry dry of all its most talented people, or perhaps join with one of its other major rivals. The latest word has it that ArtX and Nintendo are in talks to work together, perhaps under circumstances similar to those under which Nintendo would have acquired CagEnt. Unlike CagEnt, however, ArtX does not have a finished console or even half-completed chip designs to sell Nintendo, and it would be unlikely that Nintendo would be able to scrape together a reasonable system by Christmas 2000 with ArtX’s present limitations. Additionally, SGI’s recent series of strategic lawsuits against Nvidia and ArtX seem to be intended to serve as garlic and crosses to stave off any Nintendo alliance with its tastiest potential allies: Nintendo might well fear developing a new console only to find out that its core technologies or employees are depending upon infringed patents, regardless of the merits of those patents or the lawsuits.
Meanwhile, the company continues to harbor tremendous concerns for the future of the Nintendo64 platform, which appears to be sinking deeper and deeper in Japan by the day. Nintendo’s negotiations with CagEnt shed light upon the tremendous dependence the Japanese company now has upon Rare, which has been responsible for a number of the Nintendo 64’s best-looking games and at least two of the machine’s most popular—Diddy Kong Racing and Goldeneye 007. As Nintendo’s Japanese development teams have never been known for their ability to stick to release schedules, the company’s third-party rosters have remained bare and its management has remained dogmatically fixated upon silicon chips as its sole means of profit, Nintendo’s problems have set the stage for a truly interesting set of negotiations come this E3.
To sum up, readers need to understand that decisions and relationships made early in the design process of a new console can dictate a company’s standing in the industry for the following five years. Ripple effects from these decisions can be felt in a company’s bottom line can be felt for even longer. Nintendo has found itself in the unenviable position of being without an established partner and with the clock ticking down. If Nintendo should choose to go with ArtX (assuming it’s able to fight off SGI’s lawsuit), it will need to complete a chip design is an extremely short period of time. If it doesn’t go with ArtX, Nintendo will have to find a technology that is already suited to the console market or one that can readily be changed to suit a similar purpose. Either way, at this point the chances of Nintendo hitting its desired 2000 release with a new system are extremely slim.