I recall them needing more power, but I was under the impression that external power is they way they are heading, whether they add a dedicated psu to the package I don't know. There was even a mention of a totally external graphics solution with integrated power and cooling. I can't find the article though to double check so this may be still some way off in the future.

Most of what I've seen suggests we could be looking at 300-600 watts just for the GPU. Doug

The answer is we don't know. As there is no available DX10 hardware and both FSX and DX10 are still in the beta stages everything right now is just pure speculation - or, at best, an educated guess. That being said, I would think it prudent to not upgrade until the waters are much less muddy and we've had a chance to look at the real thing on both the software and hardware side. But, for me, the bigger issue is Vista itself, not DX10. To get DX10 capabilities you have to have Vista. And there is no way I'll go to Vista for at least another year. For any of us who aren't willing to immediately jump on the Vista bandwagon DX10 becomes a moot point regardless of the capabilities.Doug

I don't see any reason why the hardware guys would not market DX10-capable cards with miltiple price points and multiple performance levels. Maybe the initial release would be at the super-high end, with follow-on releases with XT, SE, GT, etc performance levels.scott s..

>>I haven't read this. Any source or anything that may suggest>how "enormous" we are talking about?I saw the article he is referring to...it was on one of the big hardware sites (Anand, Tom's, etc.) I think. It basically stated that it was their belief that the DX10 gpu's will require even more juice than the present gen cards do. They opined that 1000 watts will be pretty standard on high-end rigs in the near future.Anyway in my opinion, I am skeptical of these huge power requirements until I see some raw data on the gpu's themselves. I'm thinking those of us with a *quality* 500+ watt power supply will be ok.That means those of you with Antec TruPower 550's or Neo modulars, hehehe, you're out of luck.Rhett

>Yes, Mike said that they didn't have a DX10 card yet, but I>would expect that demonstration hardware should be showing up>pretty soon, with a November/December release. >>Where did you hear Nov/Dec DX10 card release? I heard later...as in 07Rhett

>Anyway in my opinion, I am skeptical of these huge power>requirements until I see some raw data on the gpu's>themselves. I'm thinking those of us with a *quality* 500+>watt power supply will be ok.>>That means those of you with Antec TruPower 550's or Neo>modulars, hehehe, you're out of luck.>>RhettAren't those contradictory statements? Everything I've read about the Antec 550 suggests that it is solid and very good.

ATi R600 details: (xbit labs)ATI Technologies Thursday disclosed some of the details concerning the code-named R600 graphics processing unit (GPU) and said the part will not only support the new features and unified shader architecture, but will also become the company

Its interesting.. I was reading somewhere that Nvidia are not developing unified shaders for Directx 10. Their card will have a fixed number of pixel, vertex and geometry shaders. It will be interesting to see who wins the race.

Hehe...it makes it really interesting for the developers, who now cannot simply optimize the software to take advantage of a unified shader model or a hybrid shader model. Software developers will have to either strike a balance in their code (which means that neither card will run to their full advantage) or have to favor one brand over the other, risking alienating the users of the competing brand. Makes for a lot of unhappy software developers, and in the end, they'll most likely balance the software. Funny thing about it is that despite gambling to find out which solution is the better one for DX10, the software will probably end up running about equally on either card...because it will be written that way.Why don't people ever think these things through?-Ivan

Maybe Nvidia found out that its beter(cheaper?) to have 64 of each then 'just' 64 unified shader processors. There might be some usage paterns that predict the ussualy needed shader processors per type al you have to do is supply enough of them and make them faster (specialised). The DX10 graphics card already needs the logic to switch the unified shader the hybrid card logic would simply redirect to the specialized shader. So the contest might be more like is 64 unified shaders against 64 pixel and 64 vertex shaders that don't lose time re-sheduling the shader. Technicaly elegant solution against brute force.The software is not effected because they only have to deal with the DX10 API. A hybrid card running out of a certain type of shader will be just as bad as a unified card than runs out of 'free' shaders to switch.

I suspect that it is an issue that is resolved in drivers. The actual code should be the same, the DX10 HAL should take those calls and let the driver make the calls and allow the hardware to render as it may.The R600 sounds like a monster. The x1900 is already a shader monster.

I think the point is that the code IS the same, which means that it will have to be designed for the lowest common denominator, rather than having GPU-specific codepaths to take advantage of specific features on the cards. scott s..

Do you see that in today's DX9 code? Are games optimised for the limited memory and number of shaders in Geforce 6200 TC or ATI X1300 cards?Of course not! When we have the DX10 hardware there will be budget cards that have less unified shaders. It makes sense. There is a market for budget cards. The only thing unified shaders change is the efficiency with which the available hardware is used.The biggest problem I see is that DX10 will have to create its own market. The only way that will happen is when they get a large userbase. You can only join the DX10 party by either upgrading current top of the line hardware with a DX10 Graphics card and Vista OR you buy a whole new Vista PC with top spec hardware.I think it will be at least the 2nd Quarter of 2007 before all this is even available. Much will depend on the marketshare that Vista + DX10 hardware gets after that but even if the new hardware and OS are succesfull it will be the end of 2007 before there is a commercially significant market.Even the first Voodoo 3Dfx card had it easier to gain market share because it could be added to any PC and its API worked on all OS platforms.

go here....its an artists impression of what Aces believe they can do with DX10...http://www.fs2004.com/forums/index.php?showtopic=84381