Open alyosha-tas opened 3 years ago
Thanks for forwarding these. Many of the wilbertpol tests cover CGB-E behavior and fail on real CGB-C hardware (which is what sinamas used for testing).
We do want to cover these cases to facilitate GBP console syncing, but we may spend a bit of time first getting more organized on our testing approach (and identifying other revision-specific behavior that currently exists).
I thought the sprite ones work on all models? Well all the revision stuff is really annoying whatever the case.
Some of the LY tests fail on E models too (or at least GBP.) For example: lycint152_ly153_3_dmg08_cgb04c_out00.gbc outputs 99. I think mostly stuff around LY=153 is slightly different, not sure exactly how though.
I have a bunch of tests here that can help improve mode 3 timing:
https://github.com/alyosha-tas/Misc.-GB-Tests
They test timing for window + sprites and edge cases of scroll + sprites. Note that I tested them on GBP.
Also the scroll + sprite tests here still fail on Gambatte:
https://github.com/wilbertpol/mooneye-gb/tree/master/tests/acceptance/gpu
These are of practical importance for games and aren't just some edge case stuff that never happens.