G
Guest
Guest
Since my last post I have been thinking of some of the NON-technical challenges that the Kyro II will face:
1. Since the technology between the major cards so far has been fairly similar, we have certain performance expectations from each card. There have been some differences that took us a little by surprise, like HyperZ and T-buffer. But still most new card reviews go something like this: Well it has a higher clock speed, fewer pipelines but does more textures per pass... At this point most of us already have a pretty good idea of its performance. We then just adjust that expectation slightly for whatever minor new technology was implemented.
This mind set is not only in consumers but reviewers and developers too. Nearly every benchmark is written to test "standard" hardware, and is optimized for hardware that does things the "normal" way. This is not without reason. If new cards come out that offer much better T&L, it makes sense that we find benchmarks with millions of triangles cropping up. What we <i> won't </i> see in this situation are a lot of benchmarks emphasizing overdraw. What would be the point? If the cards do poorly with overdraw, then the applications will avoid overdraw whenever possible, so an overdraw benchmark would not prove anything meaningful.
Now along comes the Kyro line and does things COMPLETELY differently. No Z-buffer? That's impossible! Tiles? ...It is uncomfortable.
It also gives performances that are unpredictable. We shoot through one of our usual benchmarks and the Kyro comes up just short of the GF2 MX. Ah-ha! we say... Now we know where this card fits in. But on the next benchmark, one which we see as very similar to the first one, it beats the GF2 Ultra. Huh? We start to scratch our heads. More benchmarks lead to more unexpected results, and this leads to debates (like this one ) and fuzzy recommendations from reviewers and retailers. Some people decide to risk going with the new technology, but many people are looking for a clear-cut winner, and this usually means they want all the usual benchmarks dominated in all the usual ways. (Or they want the CompUSA guy to say, "Yeah man, this one kills everything else!") The Kyro II does not provide this–-it provides confusion. Buyers hate confusion. It's almost always easier to buy a card like the GF2 that, though it sometimes loses to the Kyro II, performs consistently well. Again note that the "inconsistency" of the Kyro II has to do with today's software and performance expectations, not with bad technology. But the average user doesn't see this.
So here is one uphill battle for the Kyro II: getting people, from consumers to developers to reviewers to retailers, familiar with and catering to tile based renderers, just like they already are with the technology of the last five years.
2. This is related to number one, but while that was more a mind set issue, this deals with the actual performance decreases that come from that mind set. Developers are going to tailor their applications to the hardware that is most popular. Sure, developers are in a position to know the true benefits of the Kyro II better than most consumers, and they will probably exploit those benefits whenever possible. But when it comes down to a tradeoff between performance on a Kyro II and performance on a GeForce 2, they will go with the GF every time to maximize their market. This could lead to REAL (vs. misunderstood or misplaced) performance issues with the new hardware, which would fuel the doubts of the already confused consumers.
Uphill battle number two for the Kyro II: Proving and <i> maintaining </i> its worth without the developer support that the competition gets.
Needless to say this is greatly compounded by the fact that Imagination Technologies is not yet a major PC player. Had ATI or NVIDIA released a tile-based card, I think the transition would be <i> much </i> quicker.
3. Uphill battle number three: dirty tricks from the competition. I won't spend much time on this one as it is a standard underdog challenge in the business world. NVIDIA is scrutinized, and consequently, criticized more because it is on top for the moment, but companies all over the world pull this kind of stuff every day. If you really don't like these practices then you should go gripe to your congressman or somebody else that could eventually, perhaps, make a difference against them. Currently, "moral corporation" is an oxymoron; singling out and flaming NVIDIA won't do anything except make them and everybody else work all the harder at hiding their dirty side.
In conclusion, I want to point out that over and over in history technology has won acceptance and market share due to politics, not performance. I have made my views clear already in that I don't think the Kyro II is the card to buy right now. But the card does have nice strengths, and the tile-based rendering it showcases has great potential. How are politics, advertising and the other various forms of BS going to affect this technology long term?
Comments? Flames?
Regards,
Warden
====================
A computer lets you make more mistakes faster than any invention in human history - with the possible exceptions of handguns and tequila. -Mitch Ratcliffe
1. Since the technology between the major cards so far has been fairly similar, we have certain performance expectations from each card. There have been some differences that took us a little by surprise, like HyperZ and T-buffer. But still most new card reviews go something like this: Well it has a higher clock speed, fewer pipelines but does more textures per pass... At this point most of us already have a pretty good idea of its performance. We then just adjust that expectation slightly for whatever minor new technology was implemented.
This mind set is not only in consumers but reviewers and developers too. Nearly every benchmark is written to test "standard" hardware, and is optimized for hardware that does things the "normal" way. This is not without reason. If new cards come out that offer much better T&L, it makes sense that we find benchmarks with millions of triangles cropping up. What we <i> won't </i> see in this situation are a lot of benchmarks emphasizing overdraw. What would be the point? If the cards do poorly with overdraw, then the applications will avoid overdraw whenever possible, so an overdraw benchmark would not prove anything meaningful.
Now along comes the Kyro line and does things COMPLETELY differently. No Z-buffer? That's impossible! Tiles? ...It is uncomfortable.
It also gives performances that are unpredictable. We shoot through one of our usual benchmarks and the Kyro comes up just short of the GF2 MX. Ah-ha! we say... Now we know where this card fits in. But on the next benchmark, one which we see as very similar to the first one, it beats the GF2 Ultra. Huh? We start to scratch our heads. More benchmarks lead to more unexpected results, and this leads to debates (like this one ) and fuzzy recommendations from reviewers and retailers. Some people decide to risk going with the new technology, but many people are looking for a clear-cut winner, and this usually means they want all the usual benchmarks dominated in all the usual ways. (Or they want the CompUSA guy to say, "Yeah man, this one kills everything else!") The Kyro II does not provide this–-it provides confusion. Buyers hate confusion. It's almost always easier to buy a card like the GF2 that, though it sometimes loses to the Kyro II, performs consistently well. Again note that the "inconsistency" of the Kyro II has to do with today's software and performance expectations, not with bad technology. But the average user doesn't see this.
So here is one uphill battle for the Kyro II: getting people, from consumers to developers to reviewers to retailers, familiar with and catering to tile based renderers, just like they already are with the technology of the last five years.
2. This is related to number one, but while that was more a mind set issue, this deals with the actual performance decreases that come from that mind set. Developers are going to tailor their applications to the hardware that is most popular. Sure, developers are in a position to know the true benefits of the Kyro II better than most consumers, and they will probably exploit those benefits whenever possible. But when it comes down to a tradeoff between performance on a Kyro II and performance on a GeForce 2, they will go with the GF every time to maximize their market. This could lead to REAL (vs. misunderstood or misplaced) performance issues with the new hardware, which would fuel the doubts of the already confused consumers.
Uphill battle number two for the Kyro II: Proving and <i> maintaining </i> its worth without the developer support that the competition gets.
Needless to say this is greatly compounded by the fact that Imagination Technologies is not yet a major PC player. Had ATI or NVIDIA released a tile-based card, I think the transition would be <i> much </i> quicker.
3. Uphill battle number three: dirty tricks from the competition. I won't spend much time on this one as it is a standard underdog challenge in the business world. NVIDIA is scrutinized, and consequently, criticized more because it is on top for the moment, but companies all over the world pull this kind of stuff every day. If you really don't like these practices then you should go gripe to your congressman or somebody else that could eventually, perhaps, make a difference against them. Currently, "moral corporation" is an oxymoron; singling out and flaming NVIDIA won't do anything except make them and everybody else work all the harder at hiding their dirty side.
In conclusion, I want to point out that over and over in history technology has won acceptance and market share due to politics, not performance. I have made my views clear already in that I don't think the Kyro II is the card to buy right now. But the card does have nice strengths, and the tile-based rendering it showcases has great potential. How are politics, advertising and the other various forms of BS going to affect this technology long term?
Comments? Flames?
Regards,
Warden
====================
A computer lets you make more mistakes faster than any invention in human history - with the possible exceptions of handguns and tequila. -Mitch Ratcliffe