search

MiSTer-devel / SNES_MiSTer

SNES for MiSTer
GNU General Public License v3.0
197 stars 80 forks source link

More IRQ & NMI tests from higan (for reference) #164

Open James-F2 opened 4 years ago

James-F2 commented 4 years ago

https://gitlab.com/higan/snes-test-roms/tree/master/KungFuFurby-test-ROMs

Not all tests pass on the real snes,, need to do a 20 out of 20.

James-F2 commented 4 years ago

including some previous tests from here: https://github.com/MiSTer-devel/SNES_MiSTer/issues/105

Real 3chip console GPM-02 results;

demo_irqtest (pass 19/20). <--- only test MiSTer Fails consistently demo_nmitest (pass 20/20). irq (pass 0/5) real console fails consistently. (bad test?) nmi (pass 0/5) real console fails consistently. (bad test?) test_dma (20/20) test_dmasync (20/20) test_dmavalid (20/20) test_hdma (pass 10/20, fail: 10) - (MiSTer passes consistently) test_hdmasync (pass 4/10, fail 6) - (MiSTer passes consistently) test_timing (pass 5/20, fail 15) - (MiSTer passes consistently) test_irq (pass 8/20, fail 12) - (MiSTer passes consistently) test_irq4200 (pass 20/20) test_irq4209 (pass 20/20) test_irqb (pass 20/20) test_mdrhdma (pass 20/20) test_nmi (pass 7/20, fail 13) - (MiSTer passes consistently)

As for release 20200106 MiSTer is too perfect now, except first IRQ test. :)

birdybro commented 2 years ago

test_nmi (pass 7/20, fail 13) - (MiSTer passes consistently)

Just tested and this fails now on MiSTer consistently. Tracing the regression by testing different releases and... it began failing on release 20201027. The key as to why the test no longer passes is potentially within these two commits

https://github.com/MiSTer-devel/SNES_MiSTer/commit/79e55d6a3ad7cce9eec5786f40bc9483f6741400

https://github.com/MiSTer-devel/SNES_MiSTer/commit/8e53ff1b556adda5fd42a1ea95fb711d97921104

I agree, that the 3rd and 4th tests are bad tests if they fail consistently on a real console.