Next gen graphics and Vista.
- Thread starter kyetech
- Start date
Megadrive1988
Veteran
IIRC there are some things that Xenos and Xbox360's API can do that even DX10 and hardware that supports DX10 will not be able to do. however, of course, that does not make Xenos / Xbox360 capable of everything that DX10 will be able to do either, since Xenos/360 falls short of DX10 in a number of areas.
obviously R600 will be a DX10 part but we don't know about Nvidia's G80. is G80 just a refresh of G70 and thus another NV4X architecture, or is G80 the NV50 and thus a fully
DX10 / WFG2.0 / Shader 4.0 GPU ?
obviously R600 will be a DX10 part but we don't know about Nvidia's G80. is G80 just a refresh of G70 and thus another NV4X architecture, or is G80 the NV50 and thus a fully
DX10 / WFG2.0 / Shader 4.0 GPU ?
Er, I'd put those roles in reverse. nVidia's the one who will have had an SM3 architecture out for 2.5 years by the time Vista ships. If anything nVidia's the one that is going to release early.xbdestroya said:
ATI, on the other hand, will only be about one year into the R5xx architecture, and will be keen on milking that architecture as much as they can.
It's because it was "cancelled" according to some very reliable *cough* sources....Chalnoth said:
Well, my take on it is that assuming ATI has the more solid DirectX 10 strategy shaping up at the moment, any situation in which the launch of DX10 is pushed up is beneficial to ATI and detrimental to NVidia. Conversly, if NVidia gets to ride out G70 derivatives for that much longer, so much more the gravy train for them as they gain time to formulate their internal response to the looming R600.
I see your logic, but coming from the corporate perspective, I think anything that delays DX10 is of benefit financially to NVidia - unless they have an internal architecture in the works right now that will be truly stellar.
As for NV47/48:G70 - I think that's fair enough. But at the same time that's just NVidia deciding to run a little further with a refresh and dubbing it a new architecture. I don't really see any more difference between NV47 and G70 than I would the converse of the situation; say if ATI's R420 was instead named the R395.
See, this doesn't make any sense. If nVidia waits until the launch of DX10 to release their next architecture, they'll have ridden the NV4x architecture longer than any other of theirs.xbdestroya said:
It's ATI who is at risk of financial troubles due to the launch of DX10, because they may well not have had enough time by then to recoup their R&D on the R5xx.
Of course nVidia has a new architecture in the works for DX10. They've probably already begun work on the one after that.
What you seem to be suggesting here is that after producing the NV40, nVidia said, "Oh, this is good enough. We can stop work now." That's just ludicrous.
Chalnoth said:
I disagree - ATI has already suffered financially from this generation (all you need to do is review their recent earnings), and it is in their best interests to take the battle to ground where NVidia's footing is not as sure.
I'm not saying NVidia should wait until DX10 to launch their next part, but I do say that anything that pushes that up is to ATI's advantage, as far as the liklihood of putting out the better part.
R580 might serve as a suitable bridge until such a time, and if R520 and R580 - and their derivatives - start selling like gangbusters then sure, I'm sure they'd love a delay as well. But NVidia will control the pricing power throughout the rest of this gen I believe. They've already shipped significant volume at an incredibly high price point, forced ATI to lower their prices on their top end parts to stay competetive with NVidia's concurrent price drops in the 6800GT, etc... and in theory, have some tricks up their sleeves in waiting should the R520 prove to topple the GTX. It will thus be difficult for ATI to get the ROI they originally imagined this gen regardless of how successful things end up being. They still have the key holiday season to turn things around, but back to school and the summer were lost to them.
That's exactly what I'm suggesting. There is no corporate environment in which you will find innovation and 'need to exceed' to trump the drive for profit, and if NV40 is indeed 'good enough,' well it's to NVidia's advantage to ride it out. Certainly R300 has served ATI well; do you not agree?
I have no doubt that NVidia has a DX10 part in the works as well, but that said I know more about ATI's possible strategy, and in knowing something rather than nothing (as in the case of NVidia), I have no choice but to default to a mode of thinking that suggests ATI probably is commanding the high ground on DX10 at the moment.
You don't stop work because what you have is good enough, but nor do you rush your current profitable GPU's off the fabs when the upcoming design can wait to fight another day.
Last edited by a moderator:
What you're missing is that both ATI and nVidia are technology-driven companies. They each got to the positions they are currently in in the market by beating their competitors to market with better technologies. Neither company has forgotten this, and nVidia quite recently got a rude awakening with their underperforming NV30 architecture.
Both companies have, for a long time now, worked on multiple architectures simultaneously. That is to say, when the NV30 was getting read to launch, work had already begun on the NV40.
To suggest that nVidia is somehow behind ATI in terms of producing a DX10 part just doesn't make any sense to me.
Both companies have, for a long time now, worked on multiple architectures simultaneously. That is to say, when the NV30 was getting read to launch, work had already begun on the NV40.
To suggest that nVidia is somehow behind ATI in terms of producing a DX10 part just doesn't make any sense to me.
Xmas said:
I didn't say they had incentive to rush - I said that in a situation in which they were both pushed, ATI would fare better.
The benefit to ATI is that - and I'm making an assumption here - they have the stronger performing DX10 solution right now. So in a situation where DX10 suddenly becomes 'must have,' that seems to me a situation that would suit ATI very well, if even just in terms of mindshare.
@Chalnoth: I agree that the industry is technology-driven, and the fact that graphics companies have such short product cycles compared to other chip manufacturers only accentuates the fact. But the truth is where breathing room can be had, it will be taken.
Just as AMD isn't rushing out a new architecture as long as K8 is performing favorably against Intel - whereas Intel has torn all their roadmaps up - such is it that right now there is no reason for NVidia to move on from NV40. When that reason shows itself, then they'll step it up. Same with ATI - there was no real reason to dump R300-evolved solutions, as they were/are still competetive.
Of course throughout the process internal R&D continues - no one's saying it stops. But you make your move into hardware when there's a threat on the horizon to be met, or you think that doing so will translate into more dollars earned.
As for NV40, it was my understanding that NV30 caused a massive shake-up in NVidia's internal roadmaps, and indeed NV40 as it stands now wasn't the original plan. But I get this from scattered interviews I remember, so I may be off on what I remember.
Last edited by a moderator:
Well, you just have no basis for that, because the fact is that neither company has any DX10 solution right now. What they do have are designs right now. The question is: which company has the better ideas behind their designs? I don't think there's yet any reason to believe that one company's ideas will fare better than the other's.xbdestroya said:
Chalnoth said:
You're arguing a couple of different points that dont really walk together hand in hand in the industry. The NV30 had some "techncology" that was more advanced then the R300. It doesnt make it a better, in terms of quality and speed or in the over all product.
Nvidias view out in the open is that it is not quite time for Unified yet. And that rather then making a good core thats more difficult to produce, which Unified cores are, they may as well evolve current technology until Unified architecture shows promise and reason enough to switch to it. This is not to say that arent testing and working with Unified cores, they are and have been. But they need to look at it from a standpoint of, what will be gained from it and what will the costs be. Having your core unified, does not technically mean you have an advantage. Some feel it will be the opposite.
From the responces of Kirk himself, and to keep with the tradition Nvidia has been, i personally believe their flagship core of 2006 will not be unified. But i do think they will release lower end unified parts to show they arent behind. And to learn from them, how production is effected and how the market recieves them. much the same as using the 6600s as the base for the first 110nm parts.
I think its fantastic that the R600 is problably going to be very advanced, but from a performance standpoint we have no idea what to expect.
Chalnoth said:
Well Chalnoth I can't argue with that because I don't disagree. But if I had to put money on one of the two company's and their DX10 solutions right now - at this moment - I would place my money on ATI. And that's simply for the reasons I stated on the previous page: I know something about their possible plans, whereas I know nothing of NVidia's.
So if we're making predicitons - which is what this thread is about - I'm going to make them based on my best possible guess in terms of what I know. And that information points to ATI having at least the skeleton of a DX10 solution for some time now in the R400-->R500-->R600 evolution, and NVidia a big question mark.
By the way, the interview I was thinking of with regard to NVidia, NV30 and NV40 was Dave's own from quite recently.
http://www.beyond3d.com/interviews/jhh/
First question/paragraph. The implication certainly is that there was a shake-up when NV30 launched, and NV40 was as much a quickly turned-around reaction to that as anything else.
Hrm, it seems like too much of an investment to produce two rather different architectures like that.SugarCoat said:
Which is exactly what I'm attempting to suggest. Except that just bear in mind that nVidia's parts will be in the same class as ATI's when it comes to programmability, though due to ATI going full-unified, it is likely that some of the more obscure uses of unified pipelines will perform better on ATI hardware. But I don't see any reason a priori why nVidia's approach won't be better for games for some time to come.
Similar threads
