search

jotego / jtcores

FPGA cores compatible with multiple arcade game machines and KiCAD schematics of arcade games. Working on MiSTer FPGA/Analogue Pocket
https://patreon.com/jotego
GNU General Public License v3.0
236 stars 41 forks source link

bubl: Bubble Bobble - falls out of sync with PCB #221

Closed LordBarker80 closed 1 year ago

LordBarker80 commented 1 year ago

I ran the PCB and core next to each other on two almost identical cabs several times. From cold boot around 3.30 minutes in on the demo run through, level 22, the blue level with the 4 pink monsters, the core desyncs from the original hardware. I'm running the same PCB set as the default Japanese rom. I have repeated the same steps with the same result.

Thanks for your continued amazing work.

jotego commented 1 year ago

That's very interesting. Let me review the core and see if there is something suspicious. I designed it a long time ago and cannot reply off the top of my head.

LordBarker80 commented 1 year ago

That's very interesting. Let me review the core and see if there is something suspicious. I designed it a long time ago and cannot reply off the top of my head.

Thanks Jotego. Happy to send a video with the issue I spotted.

jotego commented 1 year ago

Please, when you have a minute, send it. Is the core going a bit faster or slower?

jotego commented 1 year ago

Also noticed that not so long ago I updated the core because there was one thing slowing it down. I assume you're running the newest version in JTBIN.

LordBarker80 commented 1 year ago

will do. Where can I send it? I'll check speed too.

LordBarker80 commented 1 year ago

Also noticed that not so long ago I updated the core because there was one thing slowing it down. I assume you're running the newest version in JTBIN.

Yes newest build

jotego commented 1 year ago

You can either upload here if it is not too long, or to youtube or google drive and share the link

LordBarker80 commented 1 year ago

You can either upload here if it is not too long, or to youtube or google drive and share the link

Sure give me a day or two and I'll get this uploaded and check in game speed.

LordBarker80 commented 1 year ago

You can either upload here if it is not too long, or to youtube or google drive and share the link

Hi Jotego. I have made another video to show the demo sequence. It is processing the 1080p version as we speak.

I've recorded and synced the sequence three separate times and the behaviour is the same. It 'de-syncs' as the MiSTer BB core plays two levels (at least) differently in the demo sequence to the real board.

I'm unsure if this is down to the MCU behaving incorrectly/differently or something else entirely.

The speed of the game seems correct on further tests I've done. I need to re-check level 17, for a slowdown comparison. The monster patterns in-game seem correct to me, I've only done a series of quick plays up to around level 25/26 on the MiSTer core, but seems correct.

This demo sequence change seems odd. The differences start at around 3:15/3:20 (although the level before this one has some minor demo play differences too) and become obvious around 4:45 where the comparisons become different in terms of running out of sync. I'd advise watching the whole video though as you can see the first demo levels behave the same. Thanks, look forward to hearing your thoughts.

Link: https://youtu.be/Lmvx3Bco3dc

jotego commented 1 year ago

I think that the synchronization problem in the video is most likely related to ageing in the crystal oscillator in the PCB. If MiST had gone more slowly, it could have also been related to a CPU slowdown.

Just in case I have made some improvements in the core to reduce the possibility of unintended slowdowns. I will update the binaries later today.

jotego commented 1 year ago

Another accuracy improvement done in ce66884826a148476f79ad97e753d0fd06086203