Performance Issues.

[Triticum0]

Mainly saw it during door transitions to mid 30's/50's

Door.lag.mp4

and when fighting enemies mid 40's.

Fight.lag.mp4

Doesn't occur on melonds

[MysteriousAeon]

Thanks for the report. I traced the main performance paths, and the strongest current suspect is the compute renderer's workload scaling at high internal resolution.

Could you try one controlled comparison using the same door transition or combat encounter?

Test A:

• Adaptive Widescreen: On
• HD Rendering: On
• Internal Resolution: 1x
• Texture Upscale: 1x

Test B:

• Adaptive Widescreen: On
• HD Rendering: On
• Internal Resolution: 4x
• Texture Upscale: 1x

Please restart the game between runs and use the same scene/camera/shot timing if possible.

The useful result is the minimum FPS or frame time during the exact same event.

If 4x is much slower than 1x, that would strongly support the compute tile/workload path, which may also be related to #19 and melonDS #2683.

If both settings slow down similarly, I'll look more closely at Tier-3/overlay execution, texture-cache work, and CPU-side runtime overhead instead.

[Triticum0]

Seem to effect 1x and 4x similarly was unable to redo the door test as I misconfiguration obs for some or the original testing and unwilling to rerun the test at 1x.

Test A:

Door lag (Performance drop are about same on both tests)

Large fight for morph ball (Slightly better performance than 4x)

1x.Fight.mp4

Test B:

Door lag

Untitled.Project.mp4

Going back seems to hitch more.

back.mp4

Large fight for morph ball

4x.Fight.mp4

[MysteriousAeon]

I tried to reproduce the slowdown locally using the same profiling setup, but the sampled door and combat intervals both stayed at about 59.5 FPS, so I don't have a matching slow case yet.

The videos appear to be around the Celestial Archives Meditation Room, likely entering from the Helm Room. If you still have a save from around that point, could you attach it here?

A save immediately before the affected door transition would let me profile the same game state locally without asking you to repeat the performance tests.