Now, being on the bleeding edge of .25 micron technology is not exactly a walk in the park (unless you're talking midnight in downtown LA's McArthur park) but the larger issue is the tremendous difference between the 3D only console, arcade and ARC 1 PC prototype parts that had been shown (on .35 micron, 66Mhz, 3D only) and the full blown 100Mhz, 2D/3D, .25 micron PC finished products. Adding all that extra support was probably just more difficult than anticipated unless this current story is yet another air ball and VideoLogic/NEC is pulling our lower extremity in a shameful attempt to stall for time until the whopping success of Dreamcast sprays everyone with happy juice. Either way, the bottom line is that there is a really good reason that VideoLogic (and 3Dfx) chose to do a 3D only product first - IT IS EASIER!
So then, why did VideoLogic/NEC think that adding 2D, VGA, video decode, etc. onto a rather experimental 3D part was a piece of cake? Who knows? Maybe the marketing department didn't really listen to engineering. Maybe engineering overestimated their abilities or underestimated the additional features and .25 micron challenge. Maybe everyone got a bit overexcited at the Sega deal and got caught up. Maybe the PC chip got wedged between the VideoLogic Architecture responsibilities and the joint VideoLogic/NEC Product Design partnership. Stuff happens. Just take a gander at the number of chip companies whose engineers have struggled valiantly with their technology and missed the boat, the dock and the whole dang harbor over the past two years. (List deleted to avoid irate email-Ed.)
The reason there is room for doubt here is that, although PVRS2 is being co-productized by VideoLogic under the massive corporate structure of NEC, it really is a bit risky as a design route to take for advanced 3D graphics. In contrast, underneath 3Dfx's wacky gamer image is the tried and tested OpenGL model architecture that Tarolli & Sellers used as a starting point for everything Voodoo. Compared to traditional renderers, the PVRS2 technology can look like very well organized flimflam.
Though it makes sense on paper and appeals to the neat freak resident in most engineers (myself included), until Dreamcast, deferred rendering hadn't really proved itself in the physical world. But there is another and larger part to that question: is Dreamcast (or Naomi) with its single platform, custom Hitachi CPU boosted gigapixel/second theoretical filtrate, 3D only PVRS2 part a true test of what happens with tile-based rendering in the real PC world of underpowered or overclocked PC CPUs, mismatched drivers, multiple input devices and envelope pushing game developers? (Guilty again). It will be fun to find out though, won't it? Bet you didn't realize that you were part of the big 3D graphics theory beta test, did you?
So anyhow, the development plan for the PC product was on track last spring and it took longer than expected to get the PC product into production because of unanticipated problems with the .25 micron. In other words, it had "teething problems"…The difficulty was getting the PC product with all those added features (2D, VGA, video decode, etc.) onto the .25 micron wafers. This is a contrast to our earlier speculation that Sega's production schedule demands had taken precedence over the PC product.
Because of the delay, VideoLogic and NEC have modified their product plan. In order to compete with a more mature market, they have split the team and parallel-developed two of the three planned PC chip products. Differences between the two are a higher degree of optimization on the chip, the addition of flat panel display (an upgrade from the original design) and external TV out support. Also a lower cost due to more efficient manufacturing/smaller silicon. The third PC product looks to be quite a ways off but should be something really special…
