Technical : Why not Vulkan API instead of DX11 ?

"

Obliviongamer91 wrote:

I'm not touching Win10

+1

"

Obliviongamer91 wrote:

I highly doubt DX11 will make this game run much better. In most situations, it already runs "fine"

maybe it will change not that much for the situations it 'runs fine' (and even less on engines that can handle even most of the 'heavy stuff') - but it probably can improve significant on lower average setups? It _will_ definitively add some more optimization potential.

(and yes, you're right, there is also some stuff that could be optimized in other - maybe more efficient - ways, but that should not prevent from shifting the graphics API to a much more actual level (and yes, as long as W7 is the most significant platform, and still a fully supported OS, shifting to a WX requirement would be a totally dead move)

PoE 2 should be made in Vulkan.

I think they have answered this before somewhere, when they said it's much easier/cheaper for them to go to dx11 instead of dx12/Vulcan.

Btw, for a pretty in-depth discussion of what the current limitations on POE and it's engine, watch the Lioneye's Watch podcast from yesterday with Jonathan as a guest, it gives a lot of details.

He talks about both dx11 improvements and future server improvements of the issues with recalculating "tick" damage.

@Gilgamesjh: thanks for the tip, I had missed that one. Interesting. Jonathan seems like a cool guy, someone I'd likely enjoy having a beer or three with. ;)

"

Gilgamesjh wrote:

Btw, for a pretty in-depth discussion of what the current limitations on POE and it's engine, watch the Lioneye's Watch podcast from yesterday with Jonathan as a guest, it gives a lot of details.

That's why I started this thread, I was listening the stream and thought it's too bad they don't spend that time to make something more modern. Vulkan offer more optimisation capabilities for the long term.

In the same kind of idea, it would be nice to optimize the particle engine with OpenCL.

Of course it cost time and money, but at the end that's the difference between a bad and a good graphic engine.

"

fkxCrockett wrote:

Of course it cost time and money, but at the end that's the difference between a bad and a good graphic engine.

It is probably more the decision of PoE staying in business and PoE going out of business because GGG spent a year developing the graphics pipeline, while not adding much else.

They explained it several times already, it boils down to: It would be too expensive, not worth it and it would lock their development resources.

EDIT: How many games do you know that changed their graphics engine after the game was already released?

"

Char1983 wrote:

EDIT: How many games do you know that changed their graphics engine after the game was already released?

At least 3 :
Battlefield 4, DICE added Mantle support after the release of the game.
Dota2 now support Vulkan.
Doom 2016 got Vulkan support after the release too.

You know, a well designed graphic engine can support multiple 3D API :
Doom 2016 support OpenGL and Vulkan.
Source engine support DirectX and OpenGL.
And more. Nothing new here, some 3D games was able to use both Glide and DirectX API a long time ago.

Programming speaking, you do some abstract interface to allow interoperability and portability of the code.

"

Char1983 wrote:

They explained it several times already, it boils down to: It would be too expensive, not worth it and it would lock their development resources.

Well, then maybe it was a bad idea to develop their own graphic engine if it cost too much or is too difficult for them.

Because fapping on particle engine because it finally use... SSE instructions (lol! but still not OpenCL or even DirectCompute) and now fapping on multi-threading and DX11 capabilities... While it was possible to use for example the Source engine (or better). Or they think they can do better, and in that case it isn't absurd to have a look at Vulkan and new gen technologies.

By the way, they already used the "too expansive" argument in the past about Lockstep.

Now compare the budget of the developers of said titles to GGGs budget.

I wonder why they developed their own engine, too, though I think I recall them commenting on that. Maybe it was a wrong decision, yes, but reverting that decision now is going to be hard. So, no Vulkan. Maybe with PoE2?

"

Char1983 wrote:

Now compare the budget of the developers of said titles to GGGs budget.

It was examples to answer your question, but they aren't my fav engines (except id Tech 6, OpenGL/Vulkan rocks).

The *state of the Art* engines are clearly RED engine 3 (The Witcher 3) and RAGE/Euphoria (GTA V). You can be sure both use GPU for particles and physics effects and for most possible parallel computations. And outside of technical considerations, they are good games in game-play term too.

Interesting thing, about 100 people worked on The Witcher 3. Looks to be about the current GGG team size.

"

Char1983 wrote:

I wonder why they developed their own engine, too, though I think I recall them commenting on that. Maybe it was a wrong decision, yes, but reverting that decision now is going to be hard. So, no Vulkan. Maybe with PoE2?

For sure PoE v3.0 will be DX11 but if they do it well, they could add some "abstract class" or layers, to be able to switch to others API, maybe for v3.5 or v4.0. But as they didn't planned anything, I did that thread to push the idea :p

"

Obliviongamer91 wrote:

I highly doubt DX11 will make this game run much better. In most situations, it already runs "fine"

DX11 will make the game run much smoother, since what DX9 does (or how it does what it does) is force a stop to rendering if ogg calls cannot be synced. Poe's problem are (inho) 75% sound/ogg issues, and 25% netcode/enthusiast slash non-professional coders designers.

DX11 does not have that limitation and can deal with oggs not "being there" just fine.