Latest FNA3D dll crashes on graphics device creation and destruction about fna3d

[flibitijibibo]

@bartwe

[bartwe]

Do you perhaps know if you have different directx11 feature_level support per gpu ?

[flibitijibibo]

Oh also @thatcosmonaut will care about the Vulkan part

[bartwe]

Ok managed to reproduce.

Two issues, the main one is that while it finds the right gpu, enumerating the outputs on that gpu doesn't produce any as those are connected to the low power gpu normally.
the other is that it uses D3D_IID_IDXGIAdapter instead of D3D_IID_IDXGIAdapter1 which i have a patch for.

The first issue i'm asking for more information from devs with more experience

[bartwe]

Ok gathered more information, and they recommend calling MonitorFromWindow , finding the IDXGIOutput from all of them with that HWINDOW and using that one.

[flibitijibibo]

Latest fnalibs has #141, this should be re-tested. (Also we still need to figure out the Vulkan crash.)

[thatcosmonaut]

Looking into Vulkan, it's a weird one because I can't repro this on any of my test cases and I also haven't touched the descriptor set cacheing system in about 9 months.

[flibitijibibo]

It's possibly just something from an early exit; hybrid native/managed debug info from the host system will probably work best here. I'm not able to make this happen either, so we'll see how the latest FNA3D works.

[flibitijibibo]

Latest FNA3D should be tested - a bunch of work went into the CreateDevice checks and what order the create calls are done.

[flibitijibibo]

Did we get to the bottom of this one? I may need to disable the modern DXGI path if this still isn't working for hybrid setups.

[bartwe]

Ad far as i'm aware this was resolved in #141

[flibitijibibo]

Got it - @YTN0, we'll need a response within the next 72 hours if this is not the case.