DirectX 12 ready?

[AnklaX]

DirectX 12 ready?

Will there be any efforts to make the game take advantage of directx 12 once it's out?

Found this link today showing some advantages of DX12 on the CPU front:
http://www.anandtech.com/show/8962/the-directx-12-performance-preview-amd-nvidia-star-swarm/3

I recently bought a r9 270 4gb and don't wish to buy another card just after about a year. And I'm sure I'll have to run the game at less than max settings which I obviously don't want to. It would help out a bunch for people in similar predicament if Witcher 3 would get post launch support for DX12.

So my question is - is it possible to upgrade the game for DX12 easily enough to make the effort worth the time? Or does DX12 work in such a way that the game will simply be able to take advantage of it when it comes out without the need for any additional work by the developers?

oh and what about mantle?

 

[frivolousam]

Too soon for DX12 and since this is nvidia game, I'm guessing no mantle.

 

[prince_of_nothing]

It's quite possible that CDPR might come out with a DX12/DX11.3 patch after the Witcher 3 releases. Tons of developers have done that in the past..

In fact, their lead engine programmer kinda sorta confirms that they are looking into doing just that in this interview.

CDPR has a strong relationship with both Microsoft and NVidia, so I'm sure they can get early access to DX12 if they wanted to. I even feel confident enough to predict that the DX12 patch will be bundled with the inevitable Witcher 3 Enhanced Edition

 

[GuyNwah]

DX12 just isn't timely yet. It won't be a marketable feature until Windows 10 is in GA. And that won't be until the game has been out 5 to 6 months.

The problem isn't with developer access. Developers have had access to DX12 for 4 months or so already.

It's product priorities. They have to have something to sell in May. It's unlikely that anybody on the Witcher 3 project has anything resembling the kind of free time evaluating, learning, advocating for, and demonstrating DX12 requires.

Maybe after release, especially if they can determine that DX12 will make the game more playable on marginal PCs. And DX12 pretty much obsoletes Mantle.

 

[frivolousam]

CDPR has a strong relationship with both Microsoft and NVidia

And from what I saw and read, Microsoft and AMD seem to have strong relationship over DX12, maybe there'll be a balance after all; amd users have been screwed over too much over the years.

 

[Gilrond-i-Virdan]

It would mean rewriting the engine in some way. If they are to do it, they can as well go with glnext as a major rewrite. I'm not sure though how exactly new APIs will be supported by current hardware. Would making such engine render it unusable on anything besides cards which will come out in the future?

And from what I saw and read, Microsoft and AMD seem to have strong relationship over DX12, maybe there'll be a balance after all; amd users have been screwed over too much over the years.

AMD is backing glnext, so do you think it would make sense for them to keep Mantle separate? I'd expect glnext to borrow a lot from Mantle and AMD eventually merging the effort to avoid duplication. At least it makes sense practically, but there isn't much info on that yet.

 

[GuyNwah]

It would mean rewriting the engine in some way. If they are to do it, they can as well go with glnext as a major rewrite. I'm not sure though how exactly new APIs will be supported by current hardware. Would making such engine render it unusable on anything besides cards which will come out in the future?



AMD is backing glnext, so do you think it would make sense for them to keep Mantle separate? I'd expect glnext to borrow a lot from Mantle and AMD eventually merging the effort to avoid duplication. At least it makes sense practically, but there isn't much info on that yet.

nVidia has stated all Fermi, Kepler, and Maxwell cards will "fully" support DX12. http://blogs.nvidia.com/blog/2014/09/19/maxwell-and-dx12-delivered/

AMD has stated all GCN cards (that's 7700 and up and all more recent cards) will support DX12. http://www.eteknix.com/amd-gcn-gpus-directx-12-ready/ and other sources

We should hear more about glNext support from both companies at GDC.

 

[Gilrond-i-Virdan]

Interesting, so it sounds like new API can reuse existing hardware capabilities at least in decently recent cards. That sounds promising and means that they can be used as soon as drivers will be ready. Hopefully glnext will be able to do the same.

 

[sidspyker]

I'd expect glnext to borrow a lot from Mantle and AMD eventually merging the effort to avoid duplication. At least it makes sense practically, but there isn't much info on that yet.

Well it is, AMD already stated glNext has access to Mantle.

 

[Gilrond-i-Virdan]

@sidspyker : Yes, so I'm sure they are using it. The question is, will AMD keep Mantle going forward as a separate API? I'd guess it doesn't really make sense since with glnext Mantle will unlikely get any wide adoption outside AMD hardware so they better unite the efforts.

 

[sidspyker]

That's what I'm thinking, though they did say at one point that they want to eventually make Mantle open source, which isn't the case so far.

 

[ballowers100]

When the PC version of The Witcher 3: WIld Hunt releases for sale on PC on May 19th, 2015 it probably will not have DirectX 12, DirectX 12 releases fully in around November 2015 from what I read.

If CD Projekt RED continues doing a Enhanced Edition version for the PC version of The Witcher 3: Wild Hunt on PC then possibly 5 months later the PC version of The Witcher 3: Wild Hunt Enhanced Edition can get DirectX 12 or maybe sometime in 2016.

CD Projekt RED's top priority right now is just to finish fixing bugs and polishing The Witcher 3: Wild Hunt to release for sale on May 19th, 2015.

 

[Gilrond-i-Virdan]

That's what I'm thinking, though they did say at one point that they want to eventually make Mantle open source, which isn't the case so far.

That's also probably an indirect indication that they plan to merge them. No point in doing all the work of opening Mantle (documentation and so on) if they are going to merge it with glnext which will be open standard anyway.

 

[dragonbird]

CD Projekt RED's top priority right now is just to finish fixing bugs and polishing The Witcher 3: Wild Hunt to release for sale on May 19th, 2015.

And then they would need to evaluate whether or not there is any significant benefit to introducing DX12 into an existing game, given the costs involved, as will all other developers. If the answer is "No", then it won't happen. They're not going to do it just because it exists.

(Incidentally, they've also said there won't be an Enhanced Edition, so "can get DirectX 12 or maybe sometime in 2016" may not be an accurate assessment).

 

[Gilrond-i-Virdan]

Then it makes more sense to rephrase the question to how easily can REDengine be adapted to any new graphical APIs? That wouldn't tie the issue to TW3 specifically and would put it in more proper context. Since the engine was planned to be released for other developers to use, question of supporting new APIs will arise at some point.

 

[Sken]

You would hope MS see DX12 Witcher 3 as an investment opportunity, an incentive to move to Win10 if there is a worthy frame rate bump.

 

[Gilrond-i-Virdan]

What's good for MS is not necessarily good for CDPR and their engine.

 

[Sken]

What's good for MS is not necessarily good for CDPR and their engine.

How so?

You would think RedEngine supporting the DX12 feature-set would be a plus for PC gamers.

 

[Gilrond-i-Virdan]

How so?

Because they have different interests? CDPR want to reach more users with their games and MS want to lock developers and users into their platform. So while MS of course would prefer them to use their API, at the same time CDPR won't be able to reuse that effort anywhere outside of those platforms. So it's a question of efficiency, especially if they'll have choices which can allow them reusing more of their efforts instead of duplicating them.

 

[GuyNwah]

Efficiency is a matter, not of the performance you can theoretically wring out of a version of your engine, but the number of licenses for the engine and games developed on that engine you can sell for the effort you are willing to put into it.

The proven market for CDPR games is Windows PC. Consoles are likely to join that category. Mac OS X and Linux have little following in CDPR's known markets and, barring a rapid delivery of glNext, lack a high-performance graphics platform. Without disparaging the future potential of those markets, their ability to make a profit in those markets at present wants tools, talent, and proof.

You just don't develop for and don't sell into a market that doesn't have the technology you presently need for your product and doesn't have the buyers you presently need to make a reliable forecast of sufficient profit.

These things will change as the Linux gaming market develops. But it is still sufficiently undeveloped that six and seven-figure investments are difficult to justify.