Preventing DirectX 10 on XP is something I strongly support. It will eventually, in the long run, cause a much needed break in the cycle. At some point you MUST move on or you'll stagnant development. It took DirectX 9 years (2002 release, recall that) to really catch on and people forget about that constantly. I think people often believe computer development increases faster than it does in reality.
Why didnt DX 10.1 catch on?
- Thread starter suryad
- Start date
- Status
OpenGL and DX9 also work vastly differently.
There's two big reasons:
1) Compatibility... you can't guarantee that with such a new model, all existing hardware and software will still work as-is. People might accept that of a new OS, but if you introduce a patch for an existing OS that breaks stuff, you have a problem.
2) Cost of development. You're not going to invest that much in patching up an old OS.
MS wanted to introduce various other new features in the kernel/OS anyway, so why not release an all-new version of the OS while you're at it.
Again, people should just get over it.
The whole argument is retarded anyway. In theory the whole of Vista could have been packaged as a service pack for Windows 1.0... You can patch software until you're blue in the face, the sky is the limit and all that. But from both a technical and a business point-of-view it doesn't make sense to do that.
I also find it strange that people never made a big issue out of this before. Where were all you guys when DX updates for NT stopped, and you had to move to Win2k or Win98 (shudder) for newer D3D versions? Now you're making a big deal out of it, because it's the fashionable thing to do. I'm not impressed.
That's because the entire OpenGL API implementation is inside the driver. DirectX works differently in that a good chunk of it is written by Microsoft and layered on top of a lower level interface which the drivers implement (Direct3D DDI, which is part of the Windows 2000 display driver model.). This interface has been the cause of some of DirectX's problems like for instance draw call overhead and it is this interface that's been changed for Vista and DirectX10 along with the rest of the driver model.
So while you can in theory layer a lot (most?) of the functionality of DX10 on top of the older driver model, you can't do so and retain the same performance characteristics you get with the new one. You'd get DX9.1 or so as suggested above. Microsoft obviously didn't think that was worth the effort and confusion (which DX10 do you write against? How will an application written against one run on the other?), and I tend to agree.
TimothyFarrar
Regular
Yes, and nothing was stopping MS from correcting the mistakes of DX9 such that DX10 implementation wasn't crippled in OS portability by a set driver model, DX10 was set to be non-backwards compatible anyway. It is obvious they choose this path to attempt to force people to upgrade (which I understand they have to do to keep in business), but it backfired, both for developers and users. Hence DX10.1 being not important enough to warrant hardware support via NVidia.
Which brings us to DX11. Which unlike DX10 and DX10.1 should catch on by at least console developers for next generation, which will in turn actually finally force PC gamers to upgrade at least to Vista...
Dominik D
Regular
You're still making assumption that DX10 was supposed to drive Vista sales. Or in this case that the lack of DX10 on XP hindered them. I think you're wrong, but that's irrelevant. What is important is whether your claims that it's "just" a matter of performance are valid. Do you really think that there were no design changes that made it impossible (in a reasonable time obviously) to implement DX10 on XP? I'm pretty sure port wasn't cost effective and that's why there's no DX10 on XP. In other case this would mean that MS doesn't like making money, which I hope we both agree is absurd.
Unknown Soldier
Veteran
Bouncing Zabaglione Bros.
Legend
No, I think it was a marketing decision. MS decided not to implement some kind of DX10 backwards compatibility on XP because it would force gamers and early adopters to go to Vista in order to get all those spectacular DX10 games the devs would be churning out. As so often with MS, they tried to leverage one product to sell the other.
Given the choice, it turned out that a lot of people didn't go to Vista, and so devs didn't make games for DX10.
IIRC the main design change was memory virtualisation, and that got killed when Nvidia couldn't make it work, so a compatability mode for DX10 on XP was certainly doable.
As for "reasonable time", given how late Vista was, I'm sure MS ended up with enough time if they didn't have the hubris to think everyone would jump to Vista for gaming just because of DX10.
Dominik D
Regular
I still think your assumption is wrong. There are very few DX10 games today and when Vista came out there was next to 0 DX10 games announced. How would you justify this marketing stunt in front of your managers? Hunch? Hope? It doesn't work like that. With all due respect, it's all conspiracy theories to me. Marketing in every company is making stupid mistakes but in this case with no substantial argument for DX10 being widely popular around Vista launch it's IMO very awkward to assume any other than technical reason.
Besides my argument was that WDDM (driver, software) was the reason, not GPU memory virtualization that would require hardware changes (if that's what you're referring to).
Oh puh-lease... I'm sure Windows 7 has some functionality moved to Live tools to piggyback on Windows, right? MS has to be very well aware of how these practices backfire by now, don't you think?
You want me to believe that you have the complete view on what were the technology challenges in porting DX10 to XP?
Besides my argument was that WDDM (driver, software) was the reason, not GPU memory virtualization that would require hardware changes (if that's what you're referring to).
SEGA's Stormrise and EA's Battleforge have already been announced.
Dominik D
Regular
@XMAN26: DX10.1 is mostly hardware requirements update on top of DX10. The only software side change I know of is TextureCubeArray suuport and a bunch of calls for texture information polling. My guess is that they got added because hardware already supported that (a byproduct of the way nVidia/ATI designed their HW). As far as DX10/DX10.1 games go, there's an incomplete list on WP:
http://en.wikipedia.org/wiki/List_of_games_with_DirectX_10_support
http://en.wikipedia.org/wiki/List_of_games_with_DirectX_10_support
SO you got a used to be company, SEGA aint what they used to be, and another one who treats all their customers like thieves, Not something I would really brag about Dave.
Nice list, but says nothing really outside of 1 game that did DX10.1 which was removed.
Can't remember titles, but there's at least some RTS published by EA coming, something by SEGA and apparently Blizzard too (I'm not sure if there's official word on Blizz case yet, but there is on SEGA & the game published by EA)
I'm sure lack of support by Nvidia has some impact (Assassin's Creed), but it's probably impossible to quantify. Also, ATI never refused to make a SM 3.0 card. This is closer to the SM 1.4 situation.
Dominik D
Regular
@suryad: DX10 vs. DX10.1 is mostly HW requirements differences: some features optional in 10 are mandatory in 10.1. Some cross-usage of features got enabled. Some of those things became mandatory/available because available hardware supported them anyway.
@Kaotik: AFAIK Blizzard is considering DX10 support. No official word on what effects will be exclusive to DX10 and whether any of the 10.1 features will be used.
@Kaotik: AFAIK Blizzard is considering DX10 support. No official word on what effects will be exclusive to DX10 and whether any of the 10.1 features will be used.
armchair_architect
Newcomer
I've seen this claim about Nvidia killing memory virtualization a couple of times on B3D forums, but have never seen any source (reliable or otherwise) quoted for it. Got one?
- Status
Similar threads
- Deleted member 13524
- Site Feedback