Jared M. White
Jared M. White
> Good idea, but this definitely needs a bit more work IMO. I've only skimmed this, but here's a few notes: > > * An identical timestamp does not guarantee...
No, still having the same issue.
Yoinking @Sage-King's PR to intermittently work on and potentially get merged when I think enough progress has been made in other areas to satisfy the requirements for "emulator-level" rollback. Note...
Hey man! I haven't made much progress past the other draft on page tracking; I'm currently trying to get it functional within my own branch but I'm running into some...
> What kind of feedback are you seeking on this PR at this stage? Are things working correctly right now? Mostly unsure if the direction is correct, or if I...
> The logic in the savestating code looks wrong to me. How are you recovering the position of the memory on load? Yes, that part of the logic is quite...
> Have you measured what kind of performance improvement you get from this? Before thinking about making it robust, it would be useful to know how much there is to...
For reference: 1 ms is the goal for both saves and loads, with up to 2 ms being passable; project Slippi saves and loads its state in ~1 ms depending...
> In the PR comments, you've described that you're going to make savestates be deltas rather than complete snapshots of the system. This kind of logic must not be the...