Mega Man X3 crashes on title screen

[Justturngreenpls]

On the latest stable build (2025-05-25) Mega Man X3 either freezes as its fading in to the title screen, or cuts to black shortly after fading in. It works fine on the previous stable build (2024-12-28). Rom has the CRC32 hash of fa0fe671.

[littlegamer87]

Can't reproduce. Both Megaman X2 and X3 (the only games that use CX4) run as expected on latest stable, CRC of the X3 rom is identical (USA). Tested on DE-10 HDMI with 1x128MB SDRAM.

Are you on DE10 or a clone? Have you tried SA1 and/or SuperFX/GSU games? What's the output of emudetect?

[Justturngreenpls]

I'm using an original DE10 from Terasic. Emudetect displays a screen saying "real hardware". I'll include a video demonstrating what happens when the games are loaded. There's a small edit at the end to show another way Star Fox 2 will crash with a blinking yellow screen. I tried Kirby Super Star and Super Mario RPG each for around 3 minutes and neither had any issues.
All roms play perfectly on the previous stable core.

mister.mp4

[littlegamer87]

Can you test CX4, SA1 and GSU games on the last unstable cores? If necessary up until last stable. Some chips were broken in the past unstables so some failures are to be expected.

https://github.com/MiSTer-unstable-nightlies/SNES_MiSTer/releases/tag/unstable-builds

[Justturngreenpls]

All previously tested games work in the unstable build from 2025-04-05. In the build from 2025-03-05 the CX4 games both crash on a black screen after the Capcom logo, the SA1 and GSU games seem to work fine. The 2025-02-22 build boots the GSU and CX4 games to a black screen, but plays the SA1 games fine.

[littlegamer87]

Those results from 2025-02-02 are explainable, the rest are not. The only difference between stable and last unstable commit is a sys update which cannot explain a CX4 issue, and neither can version 2025-03-05. I don't think the answer to your issue is in this repository.

Unless there is another explanation (eg bad binaries, which I can't confirm either) you might be looking at early signs of hardware failure. If that is indeed the case, use esd protection and an enclosure next time.

[TheJesusFish]

Ok, I've done some testing on this and can confirm that it is an issue with some MiSTers.

Mister 1 (de10 nano, dual ram, direct video to morph4k):

• MegaMan X2 boots and gets into gameplay fine
• MegaMan X3 boots and gets into gameplay fine
• Starfox 2 boots and gets to the main menu but hangs there (music keeps playing)

Mister 2 (Mister Pi, dual ram, direct video to tink4k)

• All games load and play fine

Mister 3 (Mister Pi, dual ram, HDMI to iPad screen)

• All games load and play fine

Mister 4 (de10 nano, dual ram, HDMI to iPad screen)

• Megaman X2 will get to various states of boot (sometimes just the cap com logo, sometimes the title, sometimes it plays all the way through)
• Megaman X3 doesn't boot (not even the cap com logo)
• Starfox 2 boots sometimes but always yellow (or blue) screens before the menu

Mister 5 (de10 nano, dual ram, direct video to CRT)

• Megaman X2 is fine
• Megaman X3 is fine
• Starfox 2 crashes occasionally - cannot figure out the trigger

Mister 6 (Qmtech, single ram (soldered on), HDMI to iPad Screen)

• Almost the exact same results as Mister 4

I built a fresh core from GitHub, and none of the games boot on that core (on any system). Other games boot fine on that core.

The unstable from April 4th fixes the issue on every mister EXCEPT Mister 6 (QMtech) which still won't let me play star fox without yellow screening.

So yeah, feels like a transient issue of some sort. Reminds me of the weirdness the Jaguar core had right after grey rewrote it (before he locked down stability).

Happy to help test more.

[Toryalai1]

It works for me as well.
Some users have a tendency to try out unofficial builds/cores. Please rule out stuff like that.

Try to nuke your SD card and reinstall MiSTer software. Then try out no intro verified rom

[TheJesusFish]

My testing was done with known good roms (crc checked) with the roms stored both on a NAS and then locally to rule out anything related to cifs.

Two misters were running the unstable main, but their results differed from each other so I don't believe that to be a factor.

[TheJesusFish]

Thanks to GrayDesire on discord, I think we may have a lead on this. I think it may be heat related.

If I let the problematic misters sit off for a few hours, and then boot right into star fox 2, it boots fine everywhere.

If I run doom 2 on ao486 for 90 minutes and then try to boot star fox 2, I get a crash at some point between boot and game type selection. All the misters have different cases and cooling solutions, so there isn’t a thru line there.

If someone can let 486 run for an hour or so, and then see if they can get star fox 2 to boot, that may help us get to the bottom of this.

I feel bad pushing on this, but I can reproduce it reliably on 4 misters, so there is something here.

[littlegamer87]

I hate to bring it to you but finding odd behavior on an FPGA when it is running hot is almost certainly hardware damage. My best advice to this would be the same I said above: use esd protection and an enclosure next time.

[TheJesusFish]

I hate to bring it to you but finding odd behavior on an FPGA when it is running hot is almost certainly hardware damage. My best advice to this would be the same I said above: use esd protection and an enclosure next time.

This is on 4 misters less than a year old.

One is less than 2 weeks old. They all are properly cooled via either fans or heatsinks.

Have you tried the test I suggested?

[littlegamer87]

Just tried it out at your request, and after exceeding 90 minutes of Doom 2 demos on 486 profile I switched to Starfox 2 and defeated those incoming missiles without crashes or issues, as expected. I still stand by what I said that an fpga that is malfunctioning when hot is (almost) certainly hardware damage, and if your failures are that early, I also believe that this is likely esd damage and that you repeated the mistake on all of them. Sorry, but I still think this repository is not the right place for answers to these issues.

I do get the idea that Starfox 2 can serve as a good mister self test. There is probably not a lot of unused die space left when that game is running, and the cycle accuracy makes malfunctions more noticeable. Bringing heat into the test was a good idea. Sorry about the result.

[TheJesusFish]

Thank you for testing.

I think your hardware failure assumption is premature, but I understand how you would arrive there. Seeing as there are multiple people who can reproduce this, I'm not quite there yet. With that said, there is obviously something up. As I mentioned earlier, I don't know if you were around for the early days of the Jaguar rework, but using the same core, some games would boot on some misters but not others.

This feels more like that - especially since everything is fine on the unstable from April and the only delta between those two builds was a sys update. Going to try to build the core from a couple of different seeds and see if any of them perform better.

Thank you again for testing. The more datapoints the better.

[TheJesusFish]

@Justturngreenpls When you have a second, try this build. I haven't been able to reproduce the issue with it (it's the current core built from seed 10).

SNES10.zip

[Justturngreenpls]

@TheJesusFish All previously nonworking games are perfectly functional on the core provided.