Timeline Update:XP,DX9,DX10


Now that my blog is all over arstechnica and digg, I need to correct something that is being misconstrued in my post about XP and DX10.


XP shipped in 2001 with DX8.1.


DX9 shipped Dec 2002. At that point the DX9 bits were final and could be used to make the desktop window manager and Aero. So that is the code branch I am talking about, and that is when DX10 wasnt “baked”. Not back in the XP timeframe.


With the public demo of  GPU memory management at WinHEC 2004 in April, thats roughly 2.5 years from the end of XP to a workable DX10.


One of my points, that what engineers enlisted into and checked code into  ( the code branch ) was well beyond XP is made stronger and not diminished by me making clear this timeline.

Comments (7)

  1. Ian McPhail says:

    Phil: your blog has been the only really useful comment on the progress of SP1 and the Dx10 patch.  In truth FSInsider is not very helpful and would be improved with this sort of news and progress reports.  Have you reached a stage where you can be a little more precise when each is likely to be available.  

    SP1 downloads will test your servers I predict.  Many thanks for the info.

  2. beemer says:

    Hi Phil

    Question re DX10.  Does DX10 render more objects more efficiently than DX9?  I read somewhere that DX10 renders 1000 objects as one object vs. DX9’s one object-to-one object render ratio.  This seems high to me, but I’m not a tech guy for sure.

    Thanks

    Ed in NY

  3. Phil Taylor says:

    Yes, DX10 renders objects more efficiently. It does this in multiple ways and here are a couple:

    1)the per-Draw invocation overhead has been reduced

    2)instancing has been dramatically improved with the addition of Texture Arrays

    3)shader constant buffers aid in constant management since there is an explicit way in the API to manage groups of constants with different update frequencies

    all of this should translate into good efficiency gains. The actual ratio depends on how much you can instance and can vary based on the game.

  4. beemer says:

    Thanks Phil

    Sounds to me like DX10 is the final perf tweek we will need.  I’m going to be patiently waiting til the fall.  Maybe I can zero in on a new machine in the meantime, based on the new AMD 65 nm’s coming out this summer.  I’m stuck in Macintosh no FSX-land now.

    Thx again for the DX10 primer.

    Ed

  5. Rigor Mortis says:

    I completely understand the reasons, why DX10 cant be integrated in XP, but there is still one thing, that i dont understand:

    Vista ist the only platform for DX10 at the moment, but what do you have DX10 and all those new gaming-technics for, on an operating system, whitch is yust not adoptable on games …

    I bought Vista, caus I have a 8800 GTX and wanted to be ready for the upcoming DX10-Games, but i couldnt stand, that the performence of my system was pull down so mutch by Vista, that there was nearly no Game, whitch I could play as fast as on XP.

    I mean, DX10, a really groundbreaking technic for high-performence-gaming, is integrated on an OS, that requires SO MANY performence only for the System ….

    In this case XP is really the better platform for games.

    I know, Microsoft is really an economy-concern and always takes the most lucrative way to sell their products, but if they would think about, what makes the best RESULT and not the most GAIN, it would be perfect !!! Quality is more important, than quantity !

    I cant believe, that Microsoft is already working on the new Windows Vienna (win7), while they are selling such a buggy OS, like Vista.

    Excuse me, that i’ve been very obvious in this post, but it’s yust, how I’m thinking about it.

    Best regards

    P.Humke

    P.S.: Excuse my bad English, I#m from Germany 😉

  6. Rigor Mortis says:

    OK, i think i shouldn’t expect an answer -.-