Many people are planning to build new systems for FSX and wondering, Is DX10 worth waiting for? Given that the cost of DX10 systems seems unknown right now, what's so great about DX10 for FS? I'm guessing this question has been asked before, only I can't find it...

>Many people are planning to build new systems for FSX and>wondering, Is DX10 worth waiting for? Given that the cost of>DX10 systems seems unknown right now, what's so great about>DX10 for FS? >>I'm guessing this question has been asked before, only I can't>find it...Sigh

I think the cost is able to be reasonably calculated.Windows Vista ~ $200-300DX10 GPU (1st gen, top of the line) ~$550-600Maybe a little more or less in total cost, depending on which version of Vista you get and which GPU you go for.If you haven't already read the blurb and seen the software rendered DX10 screenshots, then find them. I think you will be impressed.Glenn

>Many people are planning to build new systems for FSX and>wondering, Is DX10 worth waiting for? Given that the cost of>DX10 systems seems unknown right now, what's so great about>DX10 for FS? >>I'm guessing this question has been asked before, only I can't>find it...I'm sure there's some links and other information out there that can be pointed to for more clarification, but this is the best summary from what I have gathered so far:For a while, the DirectX API has been based of a similar core codebase, with just incremental changes and improvements over time, not unlike how Flight Simulator has kept aspects of its core engine and improved features and functionality with upgrades, or how the Source engine that powers Half-Life 2 is built with an open infrastructure as to invite change. With DirectX 10, instead of investigating how to suck a little more juice out of the same procedures, they looked at the core of the API itself, reevaluated, and built an entirely new process for doing what it did before. The result is not only an upgrade to the "coolness" of the graphics using new shader models and things like that, but also a momumental increase in performance. For anyone who has said that "God is in the details", they understand what graphics programmers have come to realize only very recently, which is that while you can have awesome lighting and high-poly models and normal mapping and all of that, it is the little details and the fullness and number of those details that make the world life-like. Regardless of the advancements in graphics technology, the worlds we have played in have always seemed a bit barren and empty. Games have used SpeedTree technology to put brilliant looking trees and plants into their games, but it's not until Crysis that we have looked at the screen and really felt like we were in a jungle. And it's not just the detail of the trees, but it's the pure massive number of them.And that's basically what this new upgrade seems to be promising. Not just a nice jump in the shaders to make the water look nicer and the clouds puffier, but the ability to push the number of items in a scene showing that level of detail to a much higher level. Where we once had 4 highly detailed, perfectly lit objects, we now will have 20. I think they are saying somewhere actually like 6x the performance on similarly powered hardware. But beyond that, they have some new features that should really be impressive. What I'm most excited about (ignoring things like HLSL compiler stuff and programmable pipelines) is geometry shaders. There's probably great information at wikipedia about vertex and pixel shaders that give us all of our groovy effects today, but they generally work on a finished image. You take your solid geometry (this is a rough explanation, I know), texture it, create an image and then apply your effects on top of it. I think a good analogy is to taking a polaroid of something, then drawing on top of the developed picture. Geometry shading means you can write high level shaders that actually effect the primitives themselves. Instead of just putting a "wave-like" shader effect on top of a flat water primitive, you can apply a shader directly to that element and deform the primitive to give you your wave effect, actually forming high waves. Imagine instead of just the pretty view of the way we see shader 2.x water in the demo, if, when low enough, you actually see the waves rise, swell, and break on the shores. That's what we are talking about.In addition to that, there's a stream output function in the pipeline that comes after the geometry shader, which means you can feed that information about what you had done to the geometry back in before the pixel shaders, and then apply effects to them based on how they have deformed or been "shaded". I thought of this immediately when I saw the whitecaps on the waves in the artist rendition. Feed back the deformed water, see what has reached "whitecap level" and then use a pixel shader to draw the whitecap on there, all of this happening with better overall performance than current DX9 technology.Now, I don't work for any game companies, graphics programming is a hobby, so this information is coming from my own understanding of it.But man, it's gonna be cool.

Thank you, NotASenator. Sorry, Chris, if my question seemed too basic for you. What's given some of us pause (at the alt.games.microsoft.flight-sim newsgroup) is this statement from a recent interview at simflight.com:Simflight: The detailed rendition you showed of DirectX 10 looks great, but will FSX run with DX10 at basically the same frame rate as with DX9? Microsoft: We hope to achieve that, but will not know until we have hardware, and have implemented the code. http://www.simflight.com/modules.php?name=...rticle&sid=8242

>Thank you, NotASenator. Sorry, Chris, if my question seemed>too basic for you. What's given some of us pause (at the>alt.games.microsoft.flight-sim newsgroup) is this statement>from a recent interview at simflight.com:>>Simflight: The detailed rendition you showed of DirectX 10>looks great, but will FSX run with DX10 at basically the same>frame rate as with DX9? >>Microsoft: We hope to achieve that, but will not know until we>have hardware, and have implemented the code. >>http://www.simflight.com/modules.php?name=...rticle&sid=8242Well, sure, in theory, the pipeline is more efficient. But until you actually have something to run it on, you can't tell exactly how efficient it will be. In the same way, you can build a new engine for a car, but until you actually bolt the thing on and get it on the dyno, you can't tell for sure how much horsepower you'll get out of it.

>Thank you, NotASenator. Sorry, Chris, if my question seemed>too basic for you. What's given some of us pause (at the>alt.games.microsoft.flight-sim newsgroup) is this statement>from a recent interview at simflight.com:>>Simflight: The detailed rendition you showed of DirectX 10>looks great, but will FSX run with DX10 at basically the same>frame rate as with DX9? >>Microsoft: We hope to achieve that, but will not know until we>have hardware, and have implemented the code. >>http://www.simflight.com/modules.php?name=...rticle&sid=8242I'm sorry but the millionth time a question gets asked I am not very eager to answer it.

>If you haven't already read the blurb and seen the software>rendered DX10 screenshots, then find them. I think you will be>impressed.>>GlennI'll be impressed when I see the proof in the pudding; i.e. hardware rendered shots. Then I'll wait until the 2nd generation Dx10 chip is out(let all you rich guys pay for the developement cost of the 1st gen chips) Once all that is done then I'll build a Dx10 based system.Until then I'll use my current system to run both fs9 and fsx with my Dx9 card.

Problem is that although the standards have been long set for DX10, there is still a lot of work to be done in making the hardware that will be compatible with the protocols...both ATI and nVidia seem to want to add one more DX9 card to the market before they unveil the DX10 cards. Both vendors are reported to be working on a DX10 offering, but neither has gone public with any information on what they intend to have available. Despite the secrecy, some information has managed to leak out. The ATI offering should be based on thier R600 chip, set for release around the end of 2006. Preliminary specs are as follows:65nm 64 Shader pipelines (Vec4+Scalar) 32 TMU's 32 ROPs 128 Shader Operations per Cycle 800MHz Core 102.4 billion shader ops/sec 512GFLOPs for the shaders 2 Billion triangles/sec 25.6 Gpixels/Gtexels/sec 256-bit 512MB 1.8GHz GDDR4 Memory 57.6 GB/sec Bandwidth (at 1.8GHz) WGF2.0 Unified ShaderSource: http://www.megagames.com/news/html/hardwar...srevealed.shtmlNVidia is going a slightly different route with their offering in the G80 chip, but much less is known about the specs on this card, other than the fact that nVidia will not be using a unified shader architecture.All of this is fine, but what it means to FSX is that the ACES team probably doesn't have a DX10 compliant card to play with yet. (thank the magic 8 ball for that opinion) And if they do, they've only just received it, and are still finding out what they can actually do with the hardware. Although ACES can come up with an idea of what DX10 graphics will look like, based on the specifications for DX10 compliance, exactly what will actually be renderable at a reasonable speed may very well be quite different in the end. We know that performance should increase due to the legacy code handling removal and reworking of the DLL code, but we'll be adding in the new shaders into the graphics, so the whole thing (in theory) should be a net wash...break-even. Whether the theory holds when the code gets run, however...anyone's guess at this point. Hopefully, the DX10 picture should become more clear by the November-December timeframe, when DX10 hardware begins to make its debut.

>water look nicer and the clouds puffier, but the ability to>push the number of items in a scene showing that level of>detail to a much higher level. Where we once had 4 highly>detailed, perfectly lit objects, we now will have 20. This is all very nice but I wonder at the end whether it will be worth it. I tried for example move all the sliders to the right in the FSX demo and the simulator (IMHO) actually looked worse (forget performance) - cartoonish buildings and lots of them, water "over the top", etc. I quickly setlled on some intermediate settings and in my opinion it actually looks better.Michael J.

>All of this is fine, but what it means to FSX is that the ACES>team probably doesn't have a DX10 compliant card to>play with yet. (thank the magic 8 ball for that opinion)There should be absolutely no ambiguity on this point: Mike G. (tdragger) has - more than once - flat out said they don't have any DX10 hardware... ;)

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. Especially with the fact that DX10 will be such an unusual upgrade, I would suspect that the hardware vendors will be eager to show off "killer aps" for their new hardware. I would hope that the prototypes aren't too far off, unless Vista is really going to be delayed again ;)

>>water look nicer and the clouds puffier, but the ability to>>push the number of items in a scene showing that level of>>detail to a much higher level. Where we once had 4 highly>>detailed, perfectly lit objects, we now will have 20. >>This is all very nice but I wonder at the end whether it will>be worth it. I tried for example move all the sliders to the>right in the FSX demo and the simulator (IMHO) actually looked>worse (forget performance) - cartoonish buildings and lots of>them, water "over the top", etc. I quickly setlled on some>intermediate settings and in my opinion it actually>looks better.>>Michael J.>As they say, different strokes. I know this thread, and this forum as a whole, is discussing FSX, but I know I am certainly looking past that into a whole realm of games.

Then you have to add the following for !st generation DX10 cards. based on what I've read, the 1St gen cards will require an enormous power supply. If you want to upgrade your current computer, you may have to add an upgraded power supply or an addon power supply (in an available drive bay). All in all, it sems to me to be best to wait for following generations of DX10 cards. Prices should also go down as time passes.

>Then you have to add the following for !st generation DX10>cards. based on what I've read, the 1St gen cards will>require an enormous power supply. If you want to upgrade your>current computer, you may have to add an upgraded power supply>or an addon power supply (in an available drive bay). All in>all, it sems to me to be best to wait for following>generations of DX10 cards. Prices should also go down as time>passes. I haven't read this. Any source or anything that may suggest how "enormous" we are talking about?