[Regression] Small physics bug in LittleBigPlanet [NPUA80472] (#10809)

Metraberryy commented 2 years ago

Quick summary

The wall the Skulldozer busts down in the level Skulldozer falls over early. Likely reproducible in create mode but I wouldn't know how.

Details

There isn't a whole lot I can say besides what's said in the summary, but here are some videos:

Regression: https://user-images.githubusercontent.com/30480654/180102847-c1d63fa3-5ddb-4c38-96c2-9480aa9fbb2c.mp4

Previous behaviour: https://user-images.githubusercontent.com/30480654/180103126-0faa0bb0-3cde-4352-a27d-14087faf49c5.mp4

Build with regression

v0.0.18-12773-9e62ca56 https://github.com/RPCS3/rpcs3/pull/10809

Attach two log files

https://drive.google.com/file/d/1RDmLgIKd5QfYcMm2fkTwe5Jk-wD9bI9-/view?usp=sharing

System configuration

Windows 10.0.19044 Build 19044 Intel(R) Core(TM) i7-9700K CPU @ 3.60GHz NVIDIA GeForce GTX 1660 Ti Driver Version 27.21.14.6089

TorutheRedFox commented 2 years ago

joints seem to have regressed too in both LBP1 and 2 (especially pistons, they act like strings now)

TorutheRedFox commented 2 years ago

looks like that only happens with accurate xfloat, with approximate xfloat fixing it

huh

AniLeo commented 1 year ago

@Nekotekina any idea? Very weird as it works with approximate but not accurate Also, can someone retest just to make sure it's still an issue after all the fixes that happened in between?

TorutheRedFox commented 1 year ago

yeah it's weird

accurate xfloat made my pod just completely break one time too

ruvaldak commented 1 year ago

@Nekotekina any idea? Very weird as it works with approximate but not accurate Also, can someone retest just to make sure it's still an issue after all the fixes that happened in between?

Yes this is still an issue. - in fact, it may be even worse, as it's occurring on approximate xfloat now as well.

RPCS3 / rpcs3