Who will dominate the graphics market?
- Thread starter Quitch
- Start date
Bouncing Zabaglione Bros. said:
Two points for a styling recap. 8)
Ok ByteMe; I figure you're talking about XGI, deltachrome, and Intel. I think XGI & deltachrome are already showing themselves not to be much of contenders...I dunno what Intel is up to.
Bouncing Zabaglione Bros.
Legend
digitalwanderer said:
Intel is going strictly for integrated low/low-mid end. I did read something about rumours of Nvidia and Intel possibly working together on some integrated graphics solution, but this would not be the high end tehnology leader we are talking about in this thread, but a mass market, cheaper solution.
digitalwanderer said:
Intel is a posibility.
XGI is another.
S3 is another.
PowerVR is yet another.
I hope PowerVR is one of the two that will make it into higher end cards. The other one that I predict will make it is unclear to me (I think it would be XGI or S3 more towards S3).
Two of those above will make it into the high end. Of course this would be helped if nvidia continues to suck and ATI slows down. And I believe the current releases of XGI or S3 will just be the begining (just get a toehold).
S3 should scare ATI/Nvidia/Intel. Via is huge and know how to make money. They might stay in the low end for a bit (were the mass cards are sold) and then slowy climb into performance.
What are the chances of VIA buying XGI? What kind of contender would they be then? What are the chances of Intel buying ATI? What are the chances of Nvidia dieing in the next 3-5 years? PowerVR better align themselves with someone powerful or they will also die.
Bouncing Zabaglione Bros.
Legend
ByteMe said:
Nahh, these are all low-mid end at best. We've already seen XGI and S3 current offerings arn't technology leaders as per this thread.
Sure, you posit lots of "if and buts" for maybe three or four years in the future (assuming optimistically that one year equals one generation), but these are just as likely to be true for Nvidia or ATI.
More importantly not only is ATI/Nvidia *already* making high technology graphics chips, they also have the right to lots of IP that would make it extra hard for newcomers to come in and grab their part of the high tech graphics pie.
The only people with possibly deep enough pockets to compete would be the like of Microsoft or Intel, and they have shown no interest in developing their own hardware, preferring to collaborate with the big two.
The only way a smaller company could come in is if they think of something revolutionary that enables them to leapfrog over current tech approches. Bitboys showed how difficult that is to execute, epecially when Infineon stopped making embedded ram.
PowerVR has viable tech which is different enough to throw a curve ball into the market with deferred rendering, but they have always aimed at mid-range at best, and are reliant on manufacturing partners that they don't currently have.
Well, S3 and XGI do have pretty deep pockets with VIA and SiS standing behind each (respectively).
S3 seems to have a solid game plan, I'd think that they'll probably gain some momentum/marketshare in the low-mid desktop end and eventually creep back into high-end, and then of course there's the mobile market where S3 is pretty well rooted.
S3 seems to have a solid game plan, I'd think that they'll probably gain some momentum/marketshare in the low-mid desktop end and eventually creep back into high-end, and then of course there's the mobile market where S3 is pretty well rooted.
A strong partner that is willing to invest heavily is a major presupposition for their possible success in the PC standalone graphics department. Yet since it's not the only market they address I don't think absence from that particular market would signify any "death" apart from that market only.
I'd dare to speculate that their revenues in terms of lisences/royalties would be times higher from other markets than a PC graphics card. Simple logic a PC product is usually one partner; how many partners are there already for PDA/mobile and in extension how would that possibly translate into lisencing/royalty income?
Last time I heard rumours about a possible negotiation for a partnership with VIA, there were also some rumours about VIA insisting to write the drivers....errr no thank you if that should be true.
Despite that VIA has already S3 under the hood.
Overall I'd think that with a large enough partner that is willing to invest and stay, they'd manage to re-enter the PC graphics market, yet I'd think it would be only the first step. Whoever the partner I'd say that thinking of a 3rd contender with just one attempt is pretty bold IMO.
If we'd be talking about a continuing presence in that market with frequent releases, then after N time period the potential for a real 3rd contender on all fronts would be there; and I'm clearly not talking about just one generation.
Finally if there would be ever anything for sale then it would be Imagination Technologies as a whole, I doubt they'd ever chop off just PVR, especially since I have the feeling that development and patents are in fact somewhat shared between divisions. How are you going to sell one subdivision when products X,Y,Z contain let's say Metagence or Ensigma IP is beyond me.
As for the initial question, I'd say that both ATI/NV are determined to remain as competitive to each other as they currently are. There neither continious success or failure is guaranteed at all times. I guess we'll find out coming next year after 3rd party testing/analysis of future products and not some funky speculations on possible specsheets. On paper anything looks impressive...
keegdsb said:
Deep pockets don't guarantee success; high investments and determination for numerous of sacrifices do.
Albeit the connection between each case is completely different, ST Microelectronics' pockets were deeper when they bought IP from PowerVR; in fact for 3 whole generations. One general "crisis" in the market and the division with the smallest revenue got killed. Here it's important how long a deep pocket is determined and to what extend to file red numbers...
Bouncing Zabaglione Bros. said:
The first designs called for this same 500Mhz core clock with 1Ghz memory on a 128-bit memory bus. When low-k became impossible in the required timeframe and GDDR2 production was slowed down, it was then decided to reduce those clockspeeds AFAIK. However, realizing they NEEDED them to be competitive, they decided to go back to the initial plan; decreasing yields while increasing heat and costs, of course.
Also, here are some of the features they seemed to want to achieve, and failed to:
1. Letting the VS (ab)use the PS' texture lookup units, which would have resulted in VS3.0. compliancy.
2. Non-Vec4 processing (see "CineFX Inside" from 3DCenter.org)
3. Complex gatekeeper system enabling very high amount of registers to be used, but at lower speeds; the BIG problem with that, IMO, is that the non-Vec4 system didn't work. With Vec4 only, there's not enough parallelism to be able to use units sufficiently. Furthermore, the register file was way too small; in the NV35, with a doubled register file, it's getting better. But IMO, you need even double that for the technology to make anytype of sense. I assume that will be the case in the NV40; because the base technology does seem similar in the NV40...
As you say, "NVIDIA had good ideas, but they failed to implement them well". IMO, that's partly because they nearly had "too many good ideas"; in the end, they didn't have enough time to make most of them work.
Uttar
Who gives a shit who dominates it? better would be more companies with neck to neck competition. (hoping a clean fight would be bit too much asked, I am afraid. instead of dropping prices and getting better support from game developers, companies now "optimize" for benchmarks and try to control independent reviewers. R&D is already running way too fast to game developers get all new stuff in use. The business is badly overheating and at the same time amount of different kind of new games for PC is getting smaller every month. IMO, we have a time bomb ticking here and count down is already started)
well, that was my rant for today.
well, that was my rant for today.
Nappe1 said:
Getting back on topic, IMO, the dominating IHV will be the first one to go to the next step and create their own proprietary graphics engine, possibly including sound/networking/... too if they want to be competitive.
The idea is that this engine would be capable of talking directly to the hardware of the IHV creating it, while using a general API such as DirectX or OpenGL for other IHV's hardware. Probably DX so that MS doesn't get too pissed off at this IHV. That means it would still run at excellent speeds on the competition's hardware; only, it would nearly certainly run better on their own GPUs.
And no, I don't know of any IHV currently planning such a project. I'm just stating my own personal opinion of what could create a dominating force in this industry for the forseeable future. Cg failed, because it just didn't give enough of an advantage to NVIDIA and HLSL was just as good, if not better. A good, fast engine, however, could be a much more decisive force in the upcoming IHV battles than simply better hardware, IMO. Whether it'll ever happen is another question, though.
Uttar
Well I'm thinking that it's probably be a draw with the next round of products. nVIDIA will most likely be back on track with the NV40 and the R4XX will most likely be an impressive product as well.
As to the whole MS seeing the future and the R5XX core comment made earlier. You have to remember that nVIDIA and MS had a spat a while back. It isn't all about technology if both candidates are at or near parity in terms of the offering.
As to the whole MS seeing the future and the R5XX core comment made earlier. You have to remember that nVIDIA and MS had a spat a while back. It isn't all about technology if both candidates are at or near parity in terms of the offering.
And I'd bet that said engine would be free for developers. Imagine if every developer in the world suddenly had access to an extremely competitive engine for free. You'd see EVERYONE using it.Uttar said:
I know, I know. You wouldn't see Carmack using it, for example. But all of the developers that can't afford to work on an engine for 9 months before they begin work on a game would kill to have something like that.BRiT said:
What does "NIH" stand for? (Sorry, I'm a thicky.)BRiT said:
beyondhelp
Newcomer
digitalwanderer said:
NIH = Not Invented Here
Thanks.beyondhelp said:
