MiSTer FPGA Forum

Forgive me if this has been suggested or discussed before, but looking up "saves" is kind of tricky.

I was wondering if it would be feasible to allow for saves or data write to be automatically done the moment they are done (instead of having to summon the MiSTer menu) when using an external storage like an HD, since the risk of corruption and its drawbacks seem smaller than with the system's SD card.

Not to be rude, but what does looking up saves have to do with the way saves are saved with the OSD needing to be brought up?

To look up a word means to search for it in a database, in thise case, these forums.

Ah, you meant looking it up in the forums, okay, I was just kinda confused since I was combining the two.

Automated saves is not about the volatility of the microSD, it's about the save getting corrupted because it is updating and the person turns off the system.

To be more specific, games attempt to check and write to saves at times that are not necessarily just when you write the save. This is why NES RPGs for instance had the whole "hold the reset button and press the power button" thing, because if you just pressed power, then it would potentially corrupt the SRAM, but it also was a way to make you wait for the message to say all of that, and a save won't occur after that time.

If saves are automated without the OSD, then someone could very likely be in the middle of unknowingly saving when they power their off their MiSTer, and the save could get corrupted.

Oh, I thought Sorgelig had stated the current way was because he didn't want too many SD writes way back when, my bad.

LeftEmpty wrote: ↑Thu Jan 14, 2021 4:26 pm Oh, I thought Sorgelig had stated the current way was because he didn't want too many SD writes way back when, my bad.

I remember the same thing. That's why I had been thinking about some kind of SRAM solution for saving states. In my case I'd especially be interested in arcade cores saving highscores. But of course also console games that save progress to cartridge memory.

Saves getting corrupted by turning off the device mid-save is a good point though. Maybe a core should only save when you exit out of it and indicate when it is done saving.

I don't have a problem with the current system, but if it really is an issue, couldn't the cores be programmed to write a new, separate save file rather than overwriting the previous one (the old save could be automatically deleted after the new save is confirmed to be good)? That way, a corrupt save could be easily rolled back to the previous save, and only some progress would be lost rather than all progress.

Having the cores save both automatically and when the OSD is opened might be the safest bet.

I’m surprised there isn’t more drive to get this implemented. It is infinitely more likely that you’ll lose progress because you forgot to return to the OSD or because you had a power outage during a long session where you hadn’t returned to the OSD in a while, than because of shutting it off during the extremely short time it takes to write this infinitely small SRAM data. Or like others have suggested, always keep two saves in rotation?

When you look at most other retro games solutions, like Nintendo’s Virtual Console or NES/SNES mini consoles, that’s what they do: When the games save, they save. Period. You never have to worry about your save data not being saved. If they can do it, surely the MiSTer can do it too? Please?

Having to remember to do an extra manual step just to preserve your data feels broken. It should just... work.