I've been thinking about the 9-series cards a bit and oft wondered why NVidia hasn't been using many of the same features that AMD has been using for their ATI cards. But then I came up with my own little possible conclusion.
Disclaimer: I really have no idea what the hell I'm saying. It's short and to the point because I don't know a damn thing about economics and competition, so I'm sure what I say in the next few sentences requires less brain power than the sentence you're reading now lol. So without further adieu...
Looking at the early 8800GTS and GTX cards, we see cards with more memory than NVidia is putting out today (save for the GX2 and some specialty cards from random vendors). It's no secret that 512mb, while offering ample space for the average gamer, can hinder performance at higher resolutions which is THE market for high-end gaming. In other words, it's pointless to give a high-end card less than 512mb, and would make more sense to give more than that. I think NVidia knows that since the 8800GTS, GTX, and Ultra cards come with more RAM than that and perform better at higher resolutions than the new G92 cards. Everyone knows this, I don't need to remind you. Just bear with me here.
So if NVidia has made cards with more RAM, sees sweet benchmark results at higher resolutions than the competition, then why lower the standard back to 512mb? After the G92 cards were released, we saw an obvious performance problem with higher resolutions. The problem was the reduction of RAM and bandwidth. Again, I don't need to tell you.
With the 9 series cards we see no difference in RAM size or bandwidth. Speeds have gone up but that's about it. So why, if NVidia KNOWS that they would be way better off raising the RAM to 1gb @ 512bit, did they stay at 256bit 512mb? As we've seen in the past few days, the 9800GTX is the fastest card to-date (with exception of the GX2 which is not really a single card, but some may argue that point). I think it's interesting that even though the G92 8 series cards and the 9 series cards, while having half the bandwidth and RAM, still outperform ATI's best. Is NVidia just messing with them?
Now on to the thread's topic...what is NVidia's next step? It might be better asked "what SHOULD be NVidia's next step?" If a 256bit 512mb graphics card is outperforming a 512bit 1gb card with similar clock speeds, then we can conclude that NVidia has a superior architecture. Like, by a lot. It just seems efficient enough to make up for those bottlenecks. What would happen if NVidia decided to double the bandwidth and RAM and add DX10.1 support? Would we see a giant increase in performance? I suspect that we would see a card so over the top that it would absolutely blast the competition in every way.
Does anyone else here think that NVidia is just making an attempt to keep competition alive by not introducing their best technology? Or do you think they really don't understand that their cards could do better at high-resolution gaming if they would just up the RAM and bandwidth? I don't think they're that dumb. I honestly would love to see a 512bit 1gb single-card GeForce as the next big revamp.
Disclaimer: I really have no idea what the hell I'm saying. It's short and to the point because I don't know a damn thing about economics and competition, so I'm sure what I say in the next few sentences requires less brain power than the sentence you're reading now lol. So without further adieu...
Looking at the early 8800GTS and GTX cards, we see cards with more memory than NVidia is putting out today (save for the GX2 and some specialty cards from random vendors). It's no secret that 512mb, while offering ample space for the average gamer, can hinder performance at higher resolutions which is THE market for high-end gaming. In other words, it's pointless to give a high-end card less than 512mb, and would make more sense to give more than that. I think NVidia knows that since the 8800GTS, GTX, and Ultra cards come with more RAM than that and perform better at higher resolutions than the new G92 cards. Everyone knows this, I don't need to remind you. Just bear with me here.
So if NVidia has made cards with more RAM, sees sweet benchmark results at higher resolutions than the competition, then why lower the standard back to 512mb? After the G92 cards were released, we saw an obvious performance problem with higher resolutions. The problem was the reduction of RAM and bandwidth. Again, I don't need to tell you.
With the 9 series cards we see no difference in RAM size or bandwidth. Speeds have gone up but that's about it. So why, if NVidia KNOWS that they would be way better off raising the RAM to 1gb @ 512bit, did they stay at 256bit 512mb? As we've seen in the past few days, the 9800GTX is the fastest card to-date (with exception of the GX2 which is not really a single card, but some may argue that point). I think it's interesting that even though the G92 8 series cards and the 9 series cards, while having half the bandwidth and RAM, still outperform ATI's best. Is NVidia just messing with them?
Now on to the thread's topic...what is NVidia's next step? It might be better asked "what SHOULD be NVidia's next step?" If a 256bit 512mb graphics card is outperforming a 512bit 1gb card with similar clock speeds, then we can conclude that NVidia has a superior architecture. Like, by a lot. It just seems efficient enough to make up for those bottlenecks. What would happen if NVidia decided to double the bandwidth and RAM and add DX10.1 support? Would we see a giant increase in performance? I suspect that we would see a card so over the top that it would absolutely blast the competition in every way.
Does anyone else here think that NVidia is just making an attempt to keep competition alive by not introducing their best technology? Or do you think they really don't understand that their cards could do better at high-resolution gaming if they would just up the RAM and bandwidth? I don't think they're that dumb. I honestly would love to see a 512bit 1gb single-card GeForce as the next big revamp.