Here's my view:
What 3dfx needs to do:
1) Reclaim their speed crown
2) Execute
3) Catch up, feature-wise
4) Get a new PR team
Speed:
1) Release drastically improved OGL drivers SOON for the 5500 - the 5500 SHOULD trounce the MX and DDr in MDK2 and Q3. It doesn't. This has nothing to do with T&L, and everything to do with nvidia's vast OGL driver superiority.
2) Release a higher-clocked 6000 IMMEDIATELY AFTER releasing the OGL drivers - this will allow the 6000 to DEFINITIVELY outdo the GTS-U. It can't just squeek by, it MUST whup it pretty good, and it must whup the GTS-U on it's own territory - Quake3. Say what you want about it as a game, or the validity of benchmarking, etc, but the 6000 MUST BEAT THE GTS-U AT Q3 AT ALL RESOLUTIONS/COLOR DEPTHS/FSAA SETTINGS. Period.
3) Rampage MUST be faster than the NV20. If you want to beat the champ, you can't win by decision, you MUST knock it out!!! Fully optimized DX8, OGL code is a necessity, as is a nice die-shrink.
Execute:
1) Release new OGL drivers FAST - I can't emphasize this enough. It's pitiful to see a TNT2 Ultra hanging with a 5500 in 3d Studio Max. Is this a game? No, but it tells you about the sad state of affairs that is the OGL ICD.
2) Release the 6000 NO LATER THAN 2 weeks after the GTS-U hits the streets. It had better be < 166 MHz too.
3) Release Rampage EARLY NEXT YEAR, again within striking distance of the NV20's release. nVidia CANNOT be allowed to gain a foothold in the high-end market next year, plain and simple.
Catch-up:
1) The 6000 needs to have something faster than PCI66, or it will be "AGP limited" - I suspect this is the primary issue, and why the 6k is late. It has nothing to do with AGP texturing, which is woefully inadequate, it has to do with basic data transfers needing to be a lot faster than what PCI66 can provide. That is the purpose of the Intel bridge chip on there - so the system "sees" only one graphics processor, and can then go for a higher AGP speed - incidentally, if the 6k has AGPx4 support, this may FINALLY be the grpahics adapter that can give 4x an advantage
2) Rampage MUST MUST MUST MUST have AGPx8, Dot3, EMBM, high performance T&L, DX8 compliancy. There will be no excuse.
3) This generation is, without a doubt, the LAST that 3dfx will be able to "get away" without having the "latest" features. The fillrate of cards today are such that they can handle "advanced features". The TNT1 had 32-bit color, but was woefully inadequate in fillrate to provide it @ worthwhile speeds. The SDR GeForce had T&L, but no games supported it. Those days are over. 3dfx must release a FEATURE-PACKED card in Rampage. It simply CANNOT lag behind in features.
4) They MUST pay attention to the future - don't tell us what "we need as gamers". Make it happen. If for no reason other than to avoid getting their a$$es handed to them in the mainstream press.
PR :
1) For the love of God, WHERE ARE THE T-BUFFER DEMOS? Show us how cool this new stuff is. Show me why I want it in my card. Show developers why they should implement this into their games. We all know that FSAA is pretty cool, and will be a mainstay once fillrate and throughput is available, but WHAT ABOUT THE OTHER STUFF????
2) Don't ever, ever, ever tell us what we, as gamers, "need". We'll tell you what we WANT, and you WILL respond.
------------
After using the 5500 and the GTS, I realized that there isn't all that much of a discernable difference while playing games. Guess what? Reviewers in magazines and reviewers on websites and OEMs don't give a damn about how well games play. They want a feature list, and they want benchmark scores. Give them what they want, and you'll regain your place atop the 3d graphics world.
What 3dfx needs to do:
1) Reclaim their speed crown
2) Execute
3) Catch up, feature-wise
4) Get a new PR team
Speed:
1) Release drastically improved OGL drivers SOON for the 5500 - the 5500 SHOULD trounce the MX and DDr in MDK2 and Q3. It doesn't. This has nothing to do with T&L, and everything to do with nvidia's vast OGL driver superiority.
2) Release a higher-clocked 6000 IMMEDIATELY AFTER releasing the OGL drivers - this will allow the 6000 to DEFINITIVELY outdo the GTS-U. It can't just squeek by, it MUST whup it pretty good, and it must whup the GTS-U on it's own territory - Quake3. Say what you want about it as a game, or the validity of benchmarking, etc, but the 6000 MUST BEAT THE GTS-U AT Q3 AT ALL RESOLUTIONS/COLOR DEPTHS/FSAA SETTINGS. Period.
3) Rampage MUST be faster than the NV20. If you want to beat the champ, you can't win by decision, you MUST knock it out!!! Fully optimized DX8, OGL code is a necessity, as is a nice die-shrink.
Execute:
1) Release new OGL drivers FAST - I can't emphasize this enough. It's pitiful to see a TNT2 Ultra hanging with a 5500 in 3d Studio Max. Is this a game? No, but it tells you about the sad state of affairs that is the OGL ICD.
2) Release the 6000 NO LATER THAN 2 weeks after the GTS-U hits the streets. It had better be < 166 MHz too.
3) Release Rampage EARLY NEXT YEAR, again within striking distance of the NV20's release. nVidia CANNOT be allowed to gain a foothold in the high-end market next year, plain and simple.
Catch-up:
1) The 6000 needs to have something faster than PCI66, or it will be "AGP limited" - I suspect this is the primary issue, and why the 6k is late. It has nothing to do with AGP texturing, which is woefully inadequate, it has to do with basic data transfers needing to be a lot faster than what PCI66 can provide. That is the purpose of the Intel bridge chip on there - so the system "sees" only one graphics processor, and can then go for a higher AGP speed - incidentally, if the 6k has AGPx4 support, this may FINALLY be the grpahics adapter that can give 4x an advantage
2) Rampage MUST MUST MUST MUST have AGPx8, Dot3, EMBM, high performance T&L, DX8 compliancy. There will be no excuse.
3) This generation is, without a doubt, the LAST that 3dfx will be able to "get away" without having the "latest" features. The fillrate of cards today are such that they can handle "advanced features". The TNT1 had 32-bit color, but was woefully inadequate in fillrate to provide it @ worthwhile speeds. The SDR GeForce had T&L, but no games supported it. Those days are over. 3dfx must release a FEATURE-PACKED card in Rampage. It simply CANNOT lag behind in features.
4) They MUST pay attention to the future - don't tell us what "we need as gamers". Make it happen. If for no reason other than to avoid getting their a$$es handed to them in the mainstream press.
PR :
1) For the love of God, WHERE ARE THE T-BUFFER DEMOS? Show us how cool this new stuff is. Show me why I want it in my card. Show developers why they should implement this into their games. We all know that FSAA is pretty cool, and will be a mainstay once fillrate and throughput is available, but WHAT ABOUT THE OTHER STUFF????
2) Don't ever, ever, ever tell us what we, as gamers, "need". We'll tell you what we WANT, and you WILL respond.
------------
After using the 5500 and the GTS, I realized that there isn't all that much of a discernable difference while playing games. Guess what? Reviewers in magazines and reviewers on websites and OEMs don't give a damn about how well games play. They want a feature list, and they want benchmark scores. Give them what they want, and you'll regain your place atop the 3d graphics world.
im with you 100% on that...
Facebook
Twitter