> I’m looking at each card taking its own monitor so there’s less cross-talk.

Would be a great feature!

Porting to DX11 solves none of the problems we want to solve. The killer features are:

1. Complete thread independence for actual multi-core that works.
2. Zero hidden driver costs for deterministic performance.
We have a long term roadmap for the engine that requires these things.

DX11 gets us nothing – we run on the same hardware as GL and we get the same limitations on threads and performance.

Regarding games with Vulkan renderers running slower than DX11 – this isn’t super surprising – the IHVs have spent HUGE amounts of resources optimizing the DX11 drivers, often with complete knowledge of the specific hardware. If the app devs don’t do everything perfectly, they’re going to see perf loss.

There are two possibilities:
1. This trend will never change. App vendors will never be in a position to really leverage a low level API or invest the necessary development resources.

If this is the case, then both Vulkan and DX12 are mis-designed and have too low of an abstraction level, and the right thing to do would have been a higher level API (driver does memory management, transition management, bla bla bla) but with the same thread and performance guarantees.

2. This trend will change. The app developers haven’t had time to take advantage of the architecturally unique properties of DX12 or Vulkan, so they’re competing against DX11 on DX11’s turf. Once they start doing multi-core rendering and other new-API-only possible features, they’ll be able to get perf that will not be possible on DX11, and they will just have to get ‘close’ to the DX11 driver on the low level stuff.

So everyone in the industry is hoping it’s bucket 2 and not bucket 1.

In X-Plane’s case, there are two big problems we have now that I think only get fixed with a new API:
1. We need determinism – we need to be sure that 100% of slow ops are on a background thread so we don’t stutter while flying.
2. We need to be able to safely move those ops to a background thread, which requires modern threading rules.

I don’t know. I have heard some users report that having G-Sync is a win at intermediate framerates, but I have not tested it myself.

hope you guys are up the challenge so far DX12 and Vulkan ports have been pretty bad all around and running SLOWER then DX11

supporting Vulkan means YOU THE DEVS have optimize FOR EVER SINGLE GPU architecture you want to support

and Apple seems set on Metal imo it would be better to go Metal / DX11 at this point now

in fact porting to DX11 would easier and faster then Vulkan or Metal with the tools that MS provides

your going to have support 2 APIs (3 if you count continuing to support OpenGL) any way why make even more work for what could 0 or negative gain

It already does this.

I am totally with Ben and NLS on this one. Flightsimulators have been on the bleeding edge of hardware development for the last 30 years, and will be for some time. I applaud that! I want the very best experience when I spend what little time I have with this hobby, not be hobbled by users that aren´t willing to upgrade to stay with the program. No one is taking XP10 (or 9, or 8, or…) away. If you have an old system, you are limited to old games.

maybe pausing the demo timer when the sim is paused or the user is in a settings window would be a good idea

a 1050Ti a $150usd card beats ALL 200 series cards