There are different file formats and ways to organize and save the files. In example those Dreamcast Roms on RetroAchievement expect the format from Redump.org group. That is a multi file structure with .cue and .bin files. Then I guess you have to compare the checksum to the first .bin file, but I’m not entirely sure, because the linked page does not clear this up. You maybe have gotten the files in .chd format? (BTW this is the format I prefer too.) This basically converts all .cue and .bin files of a disk into a single file format. But you would need the other previously discussed format.

• http://redump.org/discs/quicksearch/shenmue/

I just tested this by comparing the checksum for “Shenmue (Europe) (En,Fr,De,Es) (Disc 1) (Track 1).bin” listed on Redump / Shenmue (Disc 1) as 9c6830264544845880e14fe64eb94193 and it matches the first Track bin file. Normally this is how you would compare I think. HOWEVER it does not match any of the hashes listed on https://retroachievements.org/game/3398/hashes . According to their documentation the Dreamcast hashes aren’t MD5, but a custom hash that is no longer compatible with MD5: https://docs.retroachievements.org/developer-docs/game-identification.html#sega

So with the methods we have here its not sufficient just to check with MD5. This needs some extra tooling that specifically made to check Dreamcast files in Redump format. No wonder you were lost.

I would assume that they’re using the hash from redump. They’re pretty much the leading group maintaining a database of hashes for accurate disc backups. Generally most hashes for old games you’ll find online will match up with either the No-Intro database for cartridge based games, or ReDump for disc based.

Anyway, for retro games the easiest way to get a reliable copy of the game is from great big romsets people upload to Archive.org

Just Google, DDG, or search using whatever is your preference for “ReDump Dreamcast Archive.org” and you should find browsable collections.

Unfortunately, that doesn’t directly help you with figuring out the MD5 mismatch, and I’m not familiar with how exactly RetroAchievements calculates it (since it sounds like they do something weird like concatenating the game executable with something else on the disc, then get the MD5 of that). But it’s usually a safe bet that romhacks and stuff like RA are looking for one of the known good dumps from ReDump or NoIntro unless they specify otherwise.

It feels pointless to try and rip original games for use in RA. Set developers are very arbitrary in what ROM they end up developing for, and back when I used RA, I felt like I had to manually look up each potential game I wanted to play to see what version they wanted. There are vague guidelines, but zero site enforcement, for what ROMs are supposed to be used for set development.

Sometimes they develop for a PAL region because PAL releases often have multi language support. Other times NSTC/J releases were preferred for having less glitches, or glitches that they consider fun and fundamental to the proper experience of the game. Same reason is used for developing/only using specific revisions of a game. And then sometimes they just end up making RA specific patched versions. Maybe the base game has a completion tracker that isn’t actually completable, so instead of treating the original bugged value as the true 100% in the vanilla game, they force you to use a modded copy. Or perhaps the the game has a exploitable glitch that makes achievement creation pointless, since you could always “cheat” them. So they mod/patch the glitch out instead of adding a glitch check into the achievement completion logic.