Pedro J. Estébanez

@jenatali, I've checked the status of SPIR-V to DXIL via NIR and it looks amazing at this point. It really looks as it will be possible to use so this...

@jenatali, pretty good points. I'll assess. Regarding subgroup ops, we are using them if supported indeed. In any case, I think I'll start the integration work regardless some features may...

~I'm indeed overhauling the approach to shader compilation-patching in another branch. However, this one should already work. Reasons for it not to work at the moment may include the following:~...

As @Calinou has pointed out, my PR #42220 may be what this proposal is about, only that it's still limited to Android. But if it's approved, I'll do it for...

The missing piece would be to implement agile input flushing on every platform. It's still only supported on Android, if I haven't missed anything.

You're right, Agile flushing improves things in the sense that you can receive and process input multiple times per rendered frame, but it doesn't go as far as spacing game...

@TCROC, that stack trace looks like predating #93706. Can you try with #93706 itself (it would still error but at least handle it more grafecully) or with #93707, that should...

I found this at https://learn.microsoft.com/en-us/windows/win32/api/d3d12/ne-d3d12-d3d12_heap_type: > `D3D12_HEAP_TYPE_UPLOAD` > [...] > Resources in this heap must be created with `D3D12_RESOURCE_STATE_GENERIC_READ` and cannot be changed away from this. The CPU address for...

> Seems like a good enough reason to change it then, apologies for the change to common then. I can't help but wonder why the debug layer accepts it on...

#93709 should do the trick. Please confirm. However, I'm a bit surprised you run into that in the editor. As far as I can tell, the ill code path can...