But, but, but...how else can I guarantee my new viddy card will be futureproof?!?Joe DeFuria said:
Nvidia/ATI Feature philosophy
- Thread starter trinibwoy
- Start date
If I'm not mistaken, it's supported in the vertex shaders
http://msdn.microsoft.com/library/d...ence/HighLevelLanguageShaders/FlowControl.asp
http://msdn.microsoft.com/library/d...ence/HighLevelLanguageShaders/FlowControl.asp
Bouncing Zabaglione Bros.
Legend
Bjorn said:Bouncing Zabaglione Bros. said:
Usually, gamers don't just stop playing games and thus, will be able to enjoy those features faster with the next generation cards/games. So the early adopters at least get something for their money
Oh yes, but then you are not using the first generation features of that card are you? You are using features like speed, AA, AF, etc that have been refined by the previous few generations. So while you get a great card for playing current games, you (as a consumer) don't get to see much use of it's new features for another year or two while the developers finish projects using those new features for which there is now a market.
Joe DeFuria
Legend
Bjorn said:I agree and disagree
As an extreme example:
Ati releases R9700. Nvidia releases NV3X with only DX8 support but 50% faster then the R9700 and better quality FSAA.
Would it then have been wrong of fans and Ati to try and evangelize those features ?
"It depends".
If the R9700 features can be shown that, when used, provides adequate performance, then they would have a valid case for evangelization.
That being said, the nVidia fans would have a valid case for saying the NV3x is the card that has more value for the consumer, because of what it offers out of the box, and what's likely to be utilized over the life-span of the card.
Bjorn said:Joe DeFuria said:
I agree and disagree
As an extreme example:
Ati releases R9700. Nvidia releases NV3X with only DX8 support but 50% faster then the R9700 and better quality FSAA.
Would it then have been wrong of fans and Ati to try and evangelize those features ?
well a good question, but it depends on the fact - 50% slower as in can barely run DX8 @1280*1024 with no AA in normal games - well than they would be jumping the gun a little. no point in features if basic speed is not there.
On the other hand good R300 performance as it was and NV30 50% faster! but DX8.... well of course it would be good to evangelise as with newer games as Far Cry R300 would surely make up the disadvantage and give better IQ while remaining completely playable. OK NV30 would get better FPS in DX8 mode but would miss on some features too... So people would choose betwen future proof and speed... but evangelisation would have it's merits. Decent current performance is the key IMHO of course.
Joe DeFuria said:
That's true.
That depends
. What's the life span of a card ? The 9700 is what, almost 2 years old now and is hardly a "dead" card by any means. It'll probably be just as fast if not faster then the next gen budget cards. And it'll run Half Life 2 (ok, Half Life 2 might not be a good example because of the "might be delayd to 2005"..) with all the bells and whistles on. Something that the imaginary NV3X wouldn't. And if all the consumers bought the iNV3X, would we have seen any games with PS2.0 support this year ?
Whats so bad for ATi, if developers use the nv4x as a development platform for some 3_0 shader "enhanced" games?
I don't think ATi and other IHVs will wait more than a year, before their 3_0 hardware ships, so this is only a small advantage for nvidia, or mainly for their driver team.
In the end, the difference between 2_0 and 3_0 isn't that big for a developer. There are some nice things you can do, but remember for game development, frames per second are an issue: longer shader req. more power and I can't imagine a game(2004/2005/2006) which uses a lot of >96 instructions shader... I think VS3_0 Geometry Instancing will be one of the most used 3_0 features, because it can boost performance, but is more work for the developer than simply switching the HLSL compiler target ...
So before shader 3_0 only games will come out, at least every IHV will have a 3_0 or 4_0 vpu...
Thomas
I don't think ATi and other IHVs will wait more than a year, before their 3_0 hardware ships, so this is only a small advantage for nvidia, or mainly for their driver team.
In the end, the difference between 2_0 and 3_0 isn't that big for a developer. There are some nice things you can do, but remember for game development, frames per second are an issue: longer shader req. more power and I can't imagine a game(2004/2005/2006) which uses a lot of >96 instructions shader... I think VS3_0 Geometry Instancing will be one of the most used 3_0 features, because it can boost performance, but is more work for the developer than simply switching the HLSL compiler target ...
So before shader 3_0 only games will come out, at least every IHV will have a 3_0 or 4_0 vpu...
Thomas
Joe DeFuria said:
1) Pull strategy, happens in every aspect and industry of business all the time, there's nothing wrong with it.
2) Combination of the end result of the pull strategy and brand loyalty, we marketers like that. Also means that the attempt to create a perception gap/POD (point of differentiation) has succeeded.
As a marketer, I have zero issues with it because in the end, if everything goes right and it has technical merit, I sell more product and overall folks stand a good chance of having a better experience.
I don't really mind the marketing and evangelicizing of new features, it's kind of the PR dept's whole job....it's just when they use misrepresentation or disinformation in doing so that I disapprove of.
The companies are gonna hype their strong points, it's not a bad thing; it's when they start lying in trying to over-hype their strong points or cover-up there weaknesses that is the bad bit.
The companies are gonna hype their strong points, it's not a bad thing; it's when they start lying in trying to over-hype their strong points or cover-up there weaknesses that is the bad bit.
digitalwanderer said:
I don't think they lie, maybe they hide some facts....
Did you ever hear a PR person saying something bad about their own products?
That's why it is called marketing -> " the commercial processes involved in promoting and selling and distributing a product or service"
The purpose of marketing isn't to inform the buyer about the good and bad facts about a product! So I can't really blame any marketing person for doing his/her job...
Thomas
Joe DeFuria
Legend
Bjorn said:That depends
Good question...which is why as long as the feature is "usable", it deserves at least some evangelizaion to (or by) consumers.
Yes, but two years later Half-Life 2 still isn't out. And while you will be able to "play it" in full quality on the 9700, if you bought one two years ago:
1) You've COULD have been playing all your games faster and in higher quality for the past two years if you bought the i-NV3x.
2) You may also be able to play Half-Life 2 50% FASTER with the iNV3x than the Radeon 9700...just not with full quality. Which is preferable?
Joe DeFuria
Legend
tazdevl said:
Doesn't mean we have to like it.
As a consumer, it has pros and cons.
Joe DeFuria said:
That's true. But the R9700 would still be a great card (which was 2X faster then the previous generation) even if our imaginary NV3X was released. So preaching about it's PS2.0 features would have been very much OK in my book.
But if the imaginary NV3X won't play Half Life 2 50% faster ?
Nah, just kidding. I would say that that depends on the performance of the R300 and the quality difference. Looking at Far Cry before and after SM2.0 then the difference is HUGE and if the R300 can pull that of with playable framerates (which i think it can) then i'd definitely choose that.
Joe DeFuria
Legend
Bjorn said:
Again, I don't have a problem with preaching it's features...as long as it's reasonable to expect that when the features are used, you get adequate performance.
Sure. I repeat again, it all depends on how playable the performance is.
Granted, the "hard thing" to do is to guess, before games are out that use the feature, if it will be playable on that card. That's why synethetic tests become important.
Think what you will, a certain IHV's PR dept has been busted in flat-out lies.tb said:
trinibwoy said:
Before the R300 ATI had a significantly lower market-share, low developer mindshare, and bad overall perception.
They badly needed a product and a strategy to change that.
1. They needed a card that has cutting edge features so that developers can try them out and experiment (like you qouted for the NV40)
2. They had to give away a LOT of free cards to developers.
3. They had to convince developers that those features are immediatly useable.
Point 3 is actually very hard to achieve as it needs two things:
a) The new features has to run at a great speed
b) Make sure the cards are selling well to users otherwise developers wont implement features that only affects a very small amount of users.
In other words they needed a card that sells well to consumers, so:
1. They needed a card that runs existing applications exceptionally well.
2. The card has to have better IQ at low performance hit.
They did all this and it was a success.
You can see they were successfull at convincing even big developers like Valve to use the new features.
They don't need to do the same with R420 - since they are in a very different position now.
They improved market share, developer mindshare and overall perception.
So even if R420 is not the first choice for developers it's okay since the card will probably sell well anyway, perception will stay good if they market it well.
And by the time R500 arrives their status with developers will still be much better than before the R300.
OTOH, nVidia is doing a very similar tactic with the NV40 like ATI did with R300 - so let's see how it works out.
It's easy it would be an R420.
And advanced features doesn't neccessarily require more advanced shader support.
If hypotetically no new features were introduced in the next 5 years - only speed improvements, you would still see a persistent improvment of game engines and titles.
And no - no game developer will use software emulation to create effects that might have a hardware support in the future.