On Wed Oct 29 13:00:36 2025 +0000, Jacek Caban wrote:
Thanks for testing. The remaining performance difference with wined3d is likely due to suboptimal memory type selection. That’s something we should be able to tune, but I’d prefer to keep it simple and safe for the first iteration. (It’s also hardware and driver-dependent, I already see performance parity on my main box.)
Yeah, that makes sense.
Thanks again!