Main > Main Forum
Donkey Kong (us set 1) slow on mame 147
<< < (4/12) > >>
gman314:
Stupid me!  How could I leave that out?  It's an AMD 64 dual core.

By the way, I just checked the speed.  It's at 100% for most of the time, but when it hiccups, it goes down to about 97% and then quickly shoots up to 103% before going back to 100%.

I know that this is a subjective question, but what is the general consensus for the minimum percentage that a game is still playable?  At what percentage is slowdown even detectable?  I was running Crusin' USA from 80 - 90% (and even at 100% for a while).  It did seem slightly slow, but still playable. 
rCadeGaming:

--- Quote from: gman314 on November 28, 2012, 08:35:01 pm ---Stupid me!  How could I leave that out?  It's an AMD 64 dual core.
--- End quote ---

Still don't know the processor speed.  Don't need to now though...


--- Quote from: gman314 on November 28, 2012, 08:35:01 pm ---By the way, I just checked the speed.  It's at 100% for most of the time, but when it hiccups, it goes down to about 97% and then quickly shoots up to 103% before going back to 100%.
--- End quote ---

Ok, I think I know what your problem is now.  The actual output refresh rate of the graphics card is slightly off from the native refresh rate of the game.  The easiest fix will be to set frameskip to auto.  This will also help with games that have trouble running at 100%, like Cruis'n.

Really, you need to read up about what all of the video settings do.  One little change can be very significant.  For example, using some of the wrong settings can add input lag.


--- Quote from: gman314 on November 28, 2012, 08:35:01 pm ---I know that this is a subjective question, but what is the general consensus for the minimum percentage that a game is still playable?  At what percentage is slowdown even detectable?  I was running Crusin' USA from 80 - 90% (and even at 100% for a while).  It did seem slightly slow, but still playable.

--- End quote ---

Everything should be running at about 100%, or you're going to feel it, and hear some audio stuttering.
rCadeGaming:
http://strategywiki.org/wiki/MAME/Configure

some more detailed info, especially on video settings here:

http://mamedev.org/source/docs/windows.txt.html
Haze:

--- Quote from: rCadeGaming on November 28, 2012, 07:18:34 pm ---Haze, you say that you don't like fruit/gambling machines.  Who does?

--- End quote ---

Quite a lot of people, at least in the UK.  They're just a different genre really, some people hate fighters, some people hate gambling games.


--- Quote from: rCadeGaming ---These things don't technically detract from the existing traditional games in MAME, but tell me this:  Were new personnel brought on to work on this stuff, or are the existing devs now diverting their time away from improving/adding new traditional games?

--- End quote ---

Well there aren't really many people working on it, you've got me mostly looking at the non-video ones, J.Wallace who did AgeMAME (which was the old gambling build) providing assistance, and Robbie doing a lot of the Video Gamblers (I did some of those in the past)  Robbie has never really worked on anything else either, so no loss to the team there.

I still do other work on MAME / MESS too, as you can see from my page, the gambling games just form part of the cycle of things to look at, they've also helped iron out some general bugs in chip implementations and helped pave the way for things like the non-video synths MESS is starting to emulate.  Really they're just another piece of the Jigsaw.


--- Quote from: rCadeGaming ---I understand that you guys do this entirely out of the goodness of your heart.  I owe you a huge debt of gratitude for the great gift of MAME.  Of course we have no right to tell you what to do.  We're just voicing an opinion about where we, the users, wish the project was headed (I am at least).

A think feedback is always a good thing (respectful feedback at least).  We're not paying customers, so it's totally up to you how much you want to consider our feedback.  Just thinking out loud here.

--- End quote ---

Right, but some of it is just spreading misinformation / giving people a false view on things.  Putting extra things in MAME isn't determinate to MAME, creating an artificial divide doesn't have any real benefits either because if some common code needs to be made more accurate those changes have to be applied to both projects anyway, even if you split them.  I would eventually like to see MAME and MESS become one *officially* hence the UME builds I offer but I see a lot of opposition from it from people who think it will degrade the core emulator in some way because people are misinformed about such things, so posts suggesting such don't help matters.

I work on emulation with the view that everything deserves to be emulated as much as anything else, if people only emulated the absolute top games, most well known hardware platforms etc. then a lot would be lost, and neglected.  To a degree that's what was happening with the Gambling / Fruit machine stuff, the people in those scenes were typically choosing *one* set of a game, and hacking it up by any means necessary to run in an emulator that seldom saw updates while tossing every other set unchecked, and mostly unsorted in a dat file.  It wasn't even remotely clear what was dumped, wasn't dumped, was a good dump, wasn't a good dump, was complete / incomplete or anything else and without somebody stepping in to start sorting them out a bit more that situation was only going to get worse as more and more piled up; sorting through 15 years of unverified dumps done to no standard was no small task.

Now, as much as I hate the games I don't think it's right that they should be lost to history like that, they have value of sorts and a big part of the reason MAME was created was to properly document things and preserve information.  Put like that, maybe you can see that I feel it would be irresponsible of me if I hadn't started doing something about the situation?  MAME was the ideal place to start doing this work because all the basics needed were already there because MAME is a generic emulation framework and has a huge number of cores and components ready to simply be plugged together.


--- Quote from: rCadeGaming ---If it were up to me, I would want to devote more resources to removing input lag from existing games, rather than adding new ones.

--- End quote ---

I'm not sure what you mean by input lag, if you mean actual input lag (rather than using it as a way of expressing performance) then MAME isn't really adding any unavoidable lag (unless you start turning on triple buffer etc.)  You're *always* going to get 1 frame of additional lag with emulation compared to real hardware, because real hardware is drawing to the screen in real time whereas under emulation you have to wait for a frame to be complete to present it, beyond that MAME isn't specifically adding anything.  You might get some other lag depending on when controllers are read by the host operating system if that mismatches when the games read the controls, but again that's not something you can really avoid, unlike the original hardware you're not wired directly to the controls, you're going through layers of an operating system.

If you mean performance in general, yes, there are some areas in which MAME struggles, even more noticeably in MESS where the systems / games tend to be more demanding on the framework, there is no easy solution to them in many cases tho, we don't intentionally make MAME slower, just attempt to make things more accurate and develop a framework which is as easy to develop with as possible, drivers which are easy to read, all while ensuring things are portable.  Unfortunately all those things have a cost and the closer you get to 'how things work' the steeper the performance requirements; Pong is probably the ultimate example of emulating something at circuit level, it crawls, and as far as the discrete games go that's the most simple of the bunch.

Of course if you're writing something to only emulate a *single* platform then you can make more assumptions, cheat a lot more, and gain performance that way, but you also end up with code you can't reuse and one of the great things about MAME is that you have a whole bunch of reusable components which are of value way beyond the scope of something to play games.

Anyway, not the place for this discussion.
rCadeGaming:

--- Quote from: Haze on November 29, 2012, 12:01:16 am ---Anyway, not the place for this discussion.

--- End quote ---

I know, can't resist replying though, this is an important issue to me.  Let me know where is a better place.  Maybe this can be moved to another thread or something.  As far as the thread hijack, I think the OP's question may be resolved now (hopefully, we'll see if that helps him).


--- Quote from: Haze on November 29, 2012, 12:01:16 am ---Well there aren't really many people working on it, you've got me mostly looking at the non-video ones, J.Wallace who did AgeMAME (which was the old gambling build) providing assistance, and Robbie doing a lot of the Video Gamblers (I did some of those in the past)  Robbie has never really worked on anything else either, so no loss to the team there.

I still do other work on MAME / MESS too, as you can see from my page, the gambling games just form part of the cycle of things to look at, they've also helped iron out some general bugs in chip implementations and helped pave the way for things like the non-video synths MESS is starting to emulate.  Really they're just another piece of the Jigsaw.
--- End quote ---

Ok, that's cool, I didn't know that.  Just wanted to ask.


--- Quote from: Haze on November 29, 2012, 12:01:16 am ---as much as I hate the games I don't think it's right that they should be lost to history like that, they have value of sorts and a big part of the reason MAME was created was to properly document things and preserve information.  Put like that, maybe you can see that I feel it would be irresponsible of me if I hadn't started doing something about the situation?
--- End quote ---

I guess if it were me, I wouldn't be concerned with losing something to history that I don't consider worth playing.  Gambling game enthusiasts have the option to preserve it if they're interested in it, but I wouldn't feel responsible to take it upon myself.  I understand your point of view though, and it's quite noble.


--- Quote from: Haze on November 29, 2012, 12:01:16 am ---Right, but some of it is just spreading misinformation / giving people a false view on things.
--- End quote ---

Yeah, you're right.  That's why I said respectful feedback, and I did ask it as a question.  I understand the huge problem with people stating these things as fact and spreading them around. 

This has been helpful to me in understanding the situation with fruit/gambling games.  Any negative feelings I had about it have been cleared up.


--- Quote from: Haze on November 29, 2012, 12:01:16 am ---I'm not sure what you mean by input lag, if you mean actual input lag (rather than using it as a way of expressing performance) then MAME isn't really adding any unavoidable lag (unless you start turning on triple buffer etc.)  You're *always* going to get 1 frame of additional lag with emulation compared to real hardware, because real hardware is drawing to the screen in real time whereas under emulation you have to wait for a frame to be complete to present it, beyond that MAME isn't specifically adding anything.
--- End quote ---

What I said to gman was mostly about the avoidable input lag from video settings (yeah, mostly triple buffer), but I think there is some input lag that could be reduced in other places.

1 frame is outstanding, and not likely to be detectable unless it's being added to the lag of a poor display or controller encoder, etc.  However, while there are many games that run this way, I think there are also a lot of others that do add a significant lag.  For example, I noticed that Radiant Silvergun has 3 or 4 frames of lag, measured using the frame-advance method.  This may have been fixed in the latest version, I need to check, but I think there are a lot of games like this, and it does begin to significantly affect gameplay.

Let me know if the frame-advance method isn't accurate.  In the past I've used a high speed camera at 240fps to measure input lag on LCD screens, using a CRT as a benchmark.  I plan to use a similar method to test my MAME setup in the future.


--- Quote from: Haze on November 29, 2012, 12:01:16 am ---we don't intentionally make MAME slower, just attempt to make things more accurate and develop a framework which is as easy to develop with as possible, drivers which are easy to read, all while ensuring things are portable.  Unfortunately all those things have a cost and the closer you get to 'how things work' the steeper the performance requirements...

Of course if you're writing something to only emulate a *single* platform then you can make more assumptions, cheat a lot more, and gain performance that way, but you also end up with code you can't reuse and one of the great things about MAME is that you have a whole bunch of reusable components which are of value way beyond the scope of something to play games.
--- End quote ---

I appreciate this very much, and I'm willing to upgrade my hardware to run things properly.  This is why I stick with the official MAME releases, aside from the changes I make when compiling.

You may have heard of ShmupMAME, which is specifically written to minimize input lag.  It's tempting, but it seems to value that above all else, using methods that could be considered cheats or hacks; it's biased toward supporting mostly that genre of games, and it's based on the pre-107 rendering engine.  I still stick with the official MAME because I appreciate the accuracy, overall quality, and continuing improvements; but I really hope it can improve in terms of input lag in an accurate and responsible manner.

I have a limited programming experience in VB and C++ and things like that.  I'm hoping that when I get my career established and settle some other personal things I'll have some free time to develop some programming skill that's applicable to MAME, and I can contribute.
Navigation
Message Index
Next page
Previous page

Go to full version