WDDM 2.0 / 2.1
- Thread starter eigers
- Start date
This topic was briefly discussed in Fermi thread.
As far as I understand, it fell by the wayside. Maybe WDDM 1 is good enough for now and 1.1 addressed the more immediate needs. Or graphics card makers' hands are full enough and they can't spare the transistors and engineers for that. Maybe Intel and AMD sent MS to GFY, because they as hell won't want to integrate-debug something like that on first incarnations of their "integrated graphics".
As far as I understand, it fell by the wayside. Maybe WDDM 1 is good enough for now and 1.1 addressed the more immediate needs. Or graphics card makers' hands are full enough and they can't spare the transistors and engineers for that. Maybe Intel and AMD sent MS to GFY, because they as hell won't want to integrate-debug something like that on first incarnations of their "integrated graphics".
I (and most others I believe) don't know what the status of the next WDDM is.
However I think that the status of scheduling and virtualization is really messy right now and I am surprised that Microsoft did not push for a true standard there yet.
I am not so sure that a standardized memory&scheduling unit would require a significant cost over the existing architectures. How hard can it be to implement something the CPU's already got in the 80's?
The IHV's enjoy too much freedom there in my opinion and the result is that they avoid things like WDDM memory management because of the fear to lose 1-2% performance to the competition. We the dev's (and the user in the end) have to pay the bill for this.
So while I don't know about a new standard, I certainly think that something should happen.
However I think that the status of scheduling and virtualization is really messy right now and I am surprised that Microsoft did not push for a true standard there yet.
I am not so sure that a standardized memory&scheduling unit would require a significant cost over the existing architectures. How hard can it be to implement something the CPU's already got in the 80's?
The IHV's enjoy too much freedom there in my opinion and the result is that they avoid things like WDDM memory management because of the fear to lose 1-2% performance to the competition. We the dev's (and the user in the end) have to pay the bill for this.
So while I don't know about a new standard, I certainly think that something should happen.
Blazkowicz
Legend
interesting, that would be something new to bring with DX12 compatible GPUs.
I wish for multiseat configurations, i.e. two to four players each with keyb/mouse/display playing and running apps concurrently on the same machine, with GPU folding running in the winter. a pipedream maybe.
we might see that kind of thing eventually but with with expensive licensing - I feel it will be restricted to quadro/fireGL or windows server solutions rather than consumer space.
though that could get some traction in the future, with cheap optical interconnects. I basically wish for a badass PC that can be used by multiple persons throughout the house.
I wish for multiseat configurations, i.e. two to four players each with keyb/mouse/display playing and running apps concurrently on the same machine, with GPU folding running in the winter. a pipedream maybe.
we might see that kind of thing eventually but with with expensive licensing - I feel it will be restricted to quadro/fireGL or windows server solutions rather than consumer space.
though that could get some traction in the future, with cheap optical interconnects. I basically wish for a badass PC that can be used by multiple persons throughout the house.
Similar threads
