So, we are all sweating. What do you think our competition is doing? nVidia had a lock on the OEM market after ATI. Matrox has a fine product, but these aren't the days when every high end computer ships with a Millenia. s3 is hanging on for dear life, and has created a chip with a VERY short life, IMHO. nVidia has surely cut the margins on TNT by a hefty percentage, and who knows if they have lost any of the potential OEM deals. You show me your card (running at 100 MHz) and I like it and agree to use it. Then we find that it is really not as fast as promised, and decide not to pay the premium for the reduced spec...
The Permidia 3 will probably be a nice chip, but with little support. So, who do you put your money on in such a war? The company with the biggest war chest...with 100M in the bank and no debt, TDFX is looking pretty good here.
As for Banshee 2 vs TNT 2, both will get faster with the shrink. Our less complex design should mean that at any die size, we should run faster than TNT (core clock). So if we are planning on going to 125MHZ at .25mu from 100 MHz, I doubt that TNT will also go to 125 (from 90). Now, if they can get a second TMU going, Banshee would destroy TNT. Destroy it. If the retail implimentation was nothing more than a single board SLI of the Twin TMU Banshee 2, which wouldn't take too much technology, only some engineering work, you could have a card nearly 50% faster than current SLI, single board, in a window, AGP, fastest 2D, etc...Then again, they could just make a nice Banshee 2 OEM part with SLI capability; then they could sell the second board as the retail add-in...though it would be PCI. Perhaps a daughterboard? |
