So the proverbial cat is out it's proverbial bag [grin]

I'm still waiting on the final details, but a few of the DirectX MVP's who've been involved with some of the early D3D11 work will likely be at the GameFest London event on August 6th. May well see you there!

Not sure if there is anyt MVPs will be at the USA and Japan events though.

Naturally, no one who knows anything is allowed to say anything publicly but I suppose it wouldn't really be breaking any rules to say it'll be worth paying attention to...


Cheers,
Jack

Ray tracing? Not specifically but I expect more features making GPGPU easier which means ray tracing if you want to.

I can see a push towards adative tesselation / displacement at the vertex level.

Quote:more features making GPGPU easier

"Compute Shader" maybe?

Quote:I can see a push towards adative tesselation / displacement at the vertex level

"new programmable and fixed function stages designed to enable powerful, flexible tessellation" maybe?

More in my journal [wink]

Quote:Are Microsoft adopting a new strategy of having more frequent new revisions of the API, only DX9 was around for a comparatively long time. DX10 & 10.1 feel like a stop gap.

Yes, they made this assertion a few years ago!

It's all part of the fixed-caps and granular feature levels. D3D9 was around for about 5 years before 10 hit the floor, but counting D3D9Ex the API will effectively have another 3-5 years at least... And really, D3D9 has been 5 iterations - fixed function (aka D3D7 compatability), Shader-1.x (aka D3D8 compatability), Shader-2.x, Shader-3.x and then D3D9Ex (WDDM behaviour). I'm sure everyone here is familiar with how much of a mess that ended up [smile]

ISTR it was at PDC'05 where SamG answered an audience question by describing more regular (e.g. annual) API changes. Not necessarily 10,11,12,13 but at least 10.1, 10.2, 10.3, 11, 11.1 etc... This being a much saner and more manageable way of handling the rapid progress of computer graphics than the old style caps system.

Quote:Yeah i find it strange how there are relatively few games around that are DX10 compatible

Its not really that surprising to me. A lot of companies will have a mature and well understood technology platform built on D3D9-era technology. They'll have incrementally improved it over the years, but it is one advantage of the old style caps system.

Ripping all that out and really taking advanage of a new API like D3D10/10.1/11 is a huge amount of work. I'd suspect that most studios going forwards will be adopting these into any new technology they develop or buy in.

Given the product lifecycle of years you should start to see more of them appearing around now and a lot more over the next year or so as those products started when Vista+D3D10 were released are now coming around to completion and RTM.


hth
Jack

Quote:Original post by Dave
Yeah i find it strange how there are relatively few games around that are DX10 compatible. It kinda makes me tempted to support DX9 as well as 10 in the project i'm working on. Are Microsoft adopting a new strategy of having more frequent new revisions of the API, only DX9 was around for a comparatively long time. DX10 & 10.1 feel like a stop gap.

I have to agree. It feels like they're getting a bit ahead of themselves, since it's not like DX10 was such a smashing success that everyone adapted it instantly (let alone made it run decent or worked with IHV's to get actually working drivers). I'm not in the know, but I got the idea the current approach is still to do D3D9 and emulate that in D3D10, tacking on an extra feature here or there. Demirug also commented on that a while back (here), saying that the problem with D3D10 is that everyone's using it as an emulation to D3D9 concepts. Seems like something to tackle before claiming acceptance and moving on.


Oh and:

Quote:"new programmable and fixed function stages designed to enable powerful, flexible tessellation"

Fixed function?

Quote:Original post by Dave

Quote:Original post by jollyjeffers
Yes, they made this assertion a few years ago!

I feel so out of touch Mr. Hoxley.

[lol] Just me geeking out and watching all the video casts of the developer conferences [cool]


Not to go off on too much of a tangent, but the whole mentality around Windows Vista can't of done the adoption of D3D10 much help. Taken in isolation, I doubt you'll find any seasoned D3D9 developer who doesn't like the new API - either from a coding point of view or features and capability.

Quote:I got the idea the current approach is still to do D3D9 and emulate that in D3D10, tacking on an extra feature here or there

Yup, would imagine that is quite likely.

Software technology in general has this strange bipolar nature of moving very fast yet moving extremely slowly. Companies release new libraries and tools all the time but customers and developers 'like what they know and know what they like'.

Also its rare that a developer will get a clean slate to work with - you'll almost always have some sort of legacy code/content/process/design to integrate with which often limits your abilities to truly utilize all the new bells and whistles of a shiny new technology.

Moving from early shader/fixed-function hybrid solutions to 'pure' shader driven architectures with D3D9+SM3 to a simple D3D9-in-D3D10 to a 'proper' D3D10 solution seems like a realistic progression. Even when the technology is all migrated over to being D3D10 its possible that some of the old content creation tools, test suites, media etc..etc.. will still be in old formats that may hinder the engines ability to feed shiny new algorithms and show off cool "new" features.


It'll therefore be interesting to see what shape the D3D11 API takes - we've seen that 10.0 and 10.1 are very similar, but how will migrating or multi-targetting work with D3D11??

Quote:Oh and:

Quote:"new programmable and fixed function stages designed to enable powerful, flexible tessellation"

Fixed function?

ID3D11Device->MakeMyGraphicsSilkySmooth(TRUE); ?

Jack

Being Vista only meant DX10 was never going to be a smashing success. IMO slow adoption is due to the number of XP only customers out there rather than any difficulty porting to DX10.

What I'm interested to know about compute shaders is how well they fit with using the GPU for traditional rendering, I hope compute shaders and their resources are fully pipelinable with conventional rendering. (And resources fully interchangeable) I see no reason why this wouldn't be the case, crossed fingers.

Personally at the moment I'm not a big fan of moving computation from the CPU to the GPU, not until GPUs have more fillrate than artists can possibly spend. (Yeah oneday? but not yet) The HD revolution is with us and fillrate is still at a premium. Sure GPGPU is very interesting and embracing it is the future so yes, I'm glad to see more support.

I don't see multiple CPUs being a replacement to the GPU, raytracing or not. I think raytracing will merely complement rasterization technology. Raytracing on the CPU or the GPU? That will be interesting to watch, both are capable but I think GPUs will always be more naturally suited.

Quote:Original post by jollyjeffers

ID3D11Device->MakeMyGraphicsSilkySmooth(TRUE); ?

Mind your NDA [wink]

Quote:It'll therefore be interesting to see what shape the D3D11 API takes - we've seen that 10.0 and 10.1 are very similar, but how will migrating or multi-targetting work with D3D11??

That's the million dollar question. If D3D11 is too dissimilar from 10, they'll have created three (3!) platforms for developers to support on Windows. I'm not qualified to comment on the quality of the new APIs and I'm sure the new features are worth it, but it doesn't strike me as a good strategy to push out that many (perhaps significantly) distinct APIs to develop on the Windows platform alone. Not to make a drama, but it could very well hurt PC gaming.

I completely agree with the "like what they know" observation and things will of course improve. But D3D10 has been around for what, 2 years now (~1 year if you count HW support and install base)? If they keep at it like this, at some point even the most progressive developer might conclude it's probably easier and certainly more profitable to port to consoles than to keep up with all the D3D versions.