Main > Main Forum
PiMame- are we finally there?
<< < (10/13) > >>
UFO:

--- Quote from: Typefighter01 on September 22, 2013, 06:58:45 pm ---
--- Quote from: Haze on September 22, 2013, 02:48:51 pm ---
--- Quote from: UFO on September 22, 2013, 01:23:32 pm ---Has anyone got any further with a Pi build?

I'd love an update is someone is looking into this!  :cheers:

--- End quote ---

where do you expect it to go?

people have shown what the Pi is capable of, running an ancient version of MAME relatively badly.  The limits are now well defined, it's not magically going to improve overnight and you're completely handcuffed with regards what improvements you can make due to the limited capabilities of the MAME from that era if you want to backport anything newer.

taking a pi-based project any further would be sinking a lot of time and money into something for very little reward.


--- End quote ---

Wow, that was a completely unnecessary response to a perfectly reasonable question :dizzy:

UFO - It looks like Shea Silverman has a new book out called "Rasberry Pi Gaming" and I also see they compiled PiMAME to work with the Xin-Mo encoder, so looks like there is progress. I picked up a Pi for my brother for his birthday as he said he was bored and needed a project. I think I will convince him to try the MAME route and see for myself.

--- End quote ---

Thanks Typefighter...

I agree - I've supported Haze through some of his Mame battles, but meh...

Thanks for the info. I'll look that book up. For me, I was given a Pi for my birthday, and have been looking into some projects using it wil my Canon 5d. There is some interesting stuff going on.

I just came across this old thread and thought people may have taken things a little further...

:cheers:
DHTech:

--- Quote from: Haze on September 23, 2013, 12:19:15 am ---and they're all IMHO terrible from what I've seen, with people making excuses for it 'oh it's fine with a bit of frameskipping' etc...

it might be 900mhz but from what I've seen real life MAME performance is much lower than you'd expect from such a system due to other factors.

it's cheap, you get what you pay for, it's not good value, just cheap.

I really wish people would stop porting old versions of MAME to crappy platforms and just accept that proper emulation needs something significantly better hardware-wise, or something that isn't MAME and has rather been developed from the ground up for one of these ARM based platforms.  It's painful seeing the old bugs and piss-poor emulations resurrected over and over again in some zombified corpse of an emulator just because people are trying to be cheap; at least if somebody made something new they could do it with current knowledge.

--- End quote ---

Looks like somebody got out from the wrong side of the bed, and from this narrow minded view, obviously has never use a Pi.
Yes Pi can run a version of mame and some of the earlier games very well, I think this is an excellent way for people looking to get in to mame and don’t have a lot of spare cash, Pi is a great way to experience mame on a budget.
Locke141:
I like the R-pie and I don't think its very good at Mame.

If someone wants to make a hand held Mame machine, and now ones going to force me to use it, good for them.
If some one wants to stick one in a full size cap, I would recommend not to.

I would not recommend the current iterations of the PI in a Mame cabe. But one day, in the not so distant future, a R-pi model will be faster then the P4 most people are using in there cabs now. 

paigeoliver:

--- Quote from: ids on September 23, 2013, 12:13:38 am ---or .106 http://blog.sheasilverman.com/2013/01/pleasant-surprise-new-version-of-advancemame-was-released/
take your pick

--- End quote ---

Oh goodness that would have to be painfully slow. Mame 106 takes 2 to 4 times the processing power that the .37b releases did (or MUCH MUCH MORE for some titles). Just did a quick speed check using Ninja Kids with Mame 37b15 and Mame .106 and running unthrottled on my desktop pc.

Mame 37b15 650 fps
Mame 106 240 fps

I didn't pick Ninja Kids for any real reason other than it was the first thing I tried that worked. in both versions with the same romset. Actual important titles are probably going to have even bigger speed decreases as I imagine the drivers for important titles have been tweaked and worked on and improved a lot more (and thus require even more CPU cycles).

Haze:
Sorry if I do come across a bit harsh in my views on this but we've spent the last 10-15 years not only adding new stuff but fixing a lot of bugs across multiple drivers.

It would be easy to argue that MAME is irrelevant these days if you simply look at what's been added and conclude that none of that is relevant to you but like I said, if you look deeper, aside from a few long standing regressions, things are much better these days than they were even in 0.106, nevermind 0.37.

even in 0.150 I've fixed bugs in Data East hardware games Funky Jet, Rohga, and Edward Randy by improving the protection emulation.

These 3 games are all games that might be within the realm of running at 'ok' speeds on a Pi (although the build I tried at a friends place was struggling with everything 68k based for some reason, so maybe not)

That's merely one example from the latest build, if you consider that there have been hundreds of releases in that period, each one sporting a bunch of similar fixes and improvements then you can see how much they add up, even if you don't notice it immediately.

Obviously if a you port 0.150 to a Pi it's not going to run at all well, so forget that, but likewise if people resurrect 0.106, again, all they're doing is reintroducing people to a whole bunch of bugs including the ones I just mentioned.  It seems for every new platform people resurrect these same old versions with the same old bugs, so really despite having new platforms we're seeing no actual progress.  That was why my original response to the question was 'where do you expect it to go?'

That's why I wish instead of porting MAME somebody would actually start their own project specifically for these lower powered platforms where they could draw the knowledge from the very latest version of MAME, but use it in something completely fresh, without the overhead, with very specific optimizations for the platforms etc. because not even OLD versions of MAME were really optimized with something like an ARM based Pi in mind.

Porting old versions of MAME is the lazy / cheap option, so it's easy to see why it happens, but it does mean you've hit the absolute limits of what the Pi can do *unless* somebody breaks out of that cycle and starts something new.  Until then you've seen what it can do, you've seen what performance you get and you're going to have to live with all the old bugs.  Maybe some fixes can be backported, the ones I've mentioned would actually be pretty easy, but the architecture has changed a lot in many areas when it comes to video fixes etc. and the core is a lot more capable these days so even things that might look like simple backports on paper might not be at all.

Personally I feel it makes MAME look bad and it makes the Pi look bad.  While in reality old MAME versions were pretty bad, and the Pi is pretty bad too, that doesn't make me feel much better about it ;-)

I'm not trying to troll here, just trying to explain the reality of the situation.  If people really love the Pi and want it to shine they should be pushing for people to do the same thing, create something new, something fresh, something based on all the *knowledge* present in the very latest version of MAME, but at the same time something with the actual platform in mind, not just another port of the same old MAME versions people have been porting to everything for the past however long... Even if you feel the Pi is the answer, PiMame isn't.

Obviously this isn't something for Mamedev to do, our goals are simply to improve the emulation, improve the knowledge that's out there and create something to last the ages with current hardware as our target (because it makes this task much, much easier) but it is pretty depressing when people completely disregard all our current findings and spread something we know is old and buggy rather than taking the initiative and using our findings to create something better.  Like I've said, to me our work is like a document, a reference to what is correct; seeing people port MAME 0.37 is like seeing a hospital open up and exclusively use 16th century medicine because they can't get the doctors, sure, it kept people alive, but barely.

Maybe it seems strange to you for an actual Mamedev telling you that Mame isn't the answer to your problems and isn't going to take you anywhere beyond what you've already got.. but that is how I see it.

So please guys, find yourself a new hero to champion, somebody to throw your support behind, somebody who can create something new that isn't MAME instead of regurgitating all these old versions with their old bugs.  That's the way forward, that's the way really low powered devices might stand some chance of becoming decent emulation boxes, so much potential is being wasted right now by sticking with MAME in these environments even if it does have more weight behind it.


Navigation
Message Index
Next page
Previous page

Go to full version