On Mon Jul 8 12:06:34 2024 +0000, Derek Lesho wrote:
Does this fix any known application that 5988 does not?
Unreal Engine games like [1] (sack boy) for example that have a race condition triggered by a slow media source initialisation. [1]: https://www.codeweavers.com/support/bugs/browse/?cmd=bug_edit;bug_id=22038#c...
So, first of all, thanks for giving the name of a specific application.
I'm not sure which is the offending video, but I tested a couple that looked like they were relevant, and with 5988 they both resolve in 0.9-1.1 ms. That's about on the level of the measurement for Rémi's proposal given in 3606, and faster than the "0-3ms" given for Windows.
As was projected, most of the time overhead in media source resolution is related to autoplugging decoders. I don't think any other source of resolution speed was ever mentioned or even hypothesized, and these results seem to suggest that simply stopping at compressed video *is* enough.