forums.marpirc.net

I've added a new build subtarget for WolfMAME, in an attempt to exclude drivers that are of no use (ie: mechanical stuff).

It would be nice if some of you, who are able to build MAME, could grab the current WolfMAME source from https://github.com/mahlemiut/wolfmame, and try it out to see if it works, and if I've missed any mechanical drivers.
A few video gambling games have been excluded where they are non-working and/or sharing a source file with mechanical drivers. Driver count should be reduced to 11,542, but I've not seen any major change in the size of the executable.

Build using: Still trying to determine if doing this at all is worthwhile, as it will take a bit of work to keep updated. Next MAME release is scheduled for next Wednesday.

Hi Barry,

Not sure there's much point in reducing drivers, as there is no impact on execution speed or (as it appears) executable size.

Too much work for little impact if you ask me. Anyway, next MAME release will have all the MESS drivers included - are you doing to exclude all of those too?

Kranser.

I'm also trying to reduce compile time for my ageing system, and disk space, which is getting harder and harder to free up space on.

If nothing comes from this, then I'll still just build the arcade subtarget, which is equivalent to what MAME was building before, as that's more than enough.

mahlemiut wrote: I'll still just build the arcade subtarget

right on ! that is what we've all been thinking or discussing in words here.
but i hear what kranser is saying as well.

Not sure there's much point in reducing drivers, as there is no impact on execution speed or (as it appears) executable size.

Too much work for little impact if you ask me.

but here is the thing thou , recent top3 HTM files are a whopping 7000k of data (bytes) bits?
where as old top 3 data sheets measured only 1000k
got one attached if anyone wants to look at it
source: an old rom CD i made.

Ok, I've built wolf162 with this subtarget. It appears to save about 20-25MB in executable size, and another ~20MB by stripping symbols (this has the effect of making stack traces useless).

It wasn't a huge amount of work to keep the LUA and .lst for the subtarget updated, but I think I'll stick with SUBTARGET=arcade from the next version, easier to manage. But if someone wants to attempt to keep it up to date, well, that's what pull requests are for.

Since 161, I've lost the ability to compile wolfmess (and probably wolfmame), both of my xp computers give a missing instructions halfway into the mingw compiler build. Not interested in getting another computer just to compile mame. Has there been another compiler release to try or just everyone in MAME dev now never trying on older hardware?

Barry, Can I have access to a wolfmess git branch to leave my (and robbertt's) 161 changes? I probably will likely be discontinuing that upkeep, especially if I can't compile it anymore.

You can build MESS with make SUBTARGET=mess

You are welcome to fork from WolfMAME (or MAME for that matter) if you wish, it's fairly easy to keep code updated using Git, once you understand how Git works (it's very different to other version control systems).

mess mostly compiles for me but during link time I get an unknown instruction on both an intel and amd machine which leads me to believe no one is compiling it on xp anymore. Forking assumes I have a login, I do not. If you make one for me and/or get me an account to access it, I'd be happy to contribute the 161 wolfmess changes to the fork.

You should only need a Github account to fork any public repository.

Unknown instructions suggests that your CPU is lacking features that the compiler or linker needs (I remember this being mentioned what is needed, I'll need to look that up again).

I don't think XP is really supported any more, but I think it will still run there.

Update: Build tools require a CPU with SSE3 support. So you'd need a pretty old CPU to get an illegal instruction error (and one that would literally take hours to compile).