Shmupmame 4.2: Lagless Mars Matrix, Strikers 1945, Galaga 88

[IseeThings]

Again and again, the arguments of (Shmup)MAME's speed compared to PCBs should either be conducted from the standpoint of complete systems' combined/average latency or not at all. It's very much pointless to tell somebody with an LCD monitor they're having unfair advantage in latency compared to PCB players. Get real, or get out.

So you're happy to see the graphics tear apart when you move? To see everything wobble like it's going to fall apart where they mix sprites in the background layers?

It's hideous! Of course it matters, I'd never ship a game like that.

You're exaggerating this beyond belief. I take it you don't actually play games well, if at all, if you're so concerned with minor wobbling noticeable on some games in some situations? Could you point at another player here who was so offended by the graphics discrepancy as to stop using ShmupMAME (any of its versions) altogether on that count?

Otherwise, maybe you shouldn't chime in with remarks as irrelevant as that? It's no better than bashing a shmup's gameplay because you don't like the character art.

It's nothing like bashing a game because of the character art. It's like bashing a game because it looks so lazy and broken and rushed that they couldn't bother to have their sprites and backgrounds in sync. It's inexcusable.

Given the number of people who have requested such issues be FIXED in the past when they've been left broken (because sprite buffers hadn't been implemented) I imagine there are just as many who prefer the emulation to be correct, not ripped to shreds and hacked to draw video graphics out of main RAM before the game has even *sent* them to the video chips.

The point was simple, people can use this build, but it should be invalid for competitive play, it isn't accurate, it isn't correct, and it looks really amateurish on YouTube vids.

[moozooh]

Why should any emulator be valid for a competition then?

[IseeThings]

Why should any emulator be valid for a competition then?

For real world records, I don't think they should.

For the purpose of MARP etc. You've got 99% of the people playing with regular legitimate versions, using a broken custom version shouldn't be valid.

For the purpose of play here, it should be clearly stated. Some people won't want the broken graphics. Does it become fair to compare the scores if you're claiming this build gives such a play advantage (which in the case of the Raizing games WILL be beyond even the possibility of the PCB in a worst-case triple-buffer turned on scenario because the hacks are drawing the video data before it even gets sent to the video chip!)

[Barrakketh]

Call me thick or a tard but how can LCD screens give more lag compared to CRT? Is it all interpollation or is there lag even at native res?

Post-processing can cause it (most TVs should have a "game mode" that disables it), but the pixels in an LCD panel also don't respond immediately. The 2-4ms response time you see on monitors with a TN panel is the gray-to-gray time. That's just the nature of the beast.

[moozooh]

For real world records, I don't think they should.

For the purpose of MARP etc. You've got 99% of the people playing with regular legitimate versions, using a broken custom version shouldn't be valid.

For the purpose of play here, it should be clearly stated. Some people won't want the broken graphics. Does it become fair to compare the scores if you're claiming this build gives such a play advantage (which in the case of the Raizing games WILL be beyond even the possibility of the PCB in a worst-case triple-buffer turned on scenario because the hacks are drawing the video data before it even gets sent to the video chip!)

For real world records, they obviously shouldn't. But, ironically, you're taking competitive play here both too seriously (when it comes to allowable limitations) and not seriously enough (when it comes to deciding what matters more between a minor graphics glitch and increase in responsiveness).

As for legitimacy, you're either being patronizing, hypocritical, or both — choose for yourself. Are any of the versions listed here good for competition despite being old and crummy and whatnot? What about 0.6x, 0.7x, and 0.8x versions that many (probably most, in fact) MARP replays are recorded with? What about that version somewhere in early hundreds where a frame of lag was artificially added to simulate the hardware "more accurately"? I'm of the opinion that the ground is so unlevel there, of all places, that your righteous act of notifying the admins of the follies of ShmupMAME, yet completely ignoring whatever other follies that have taken place in the official MAME builds and ended up on MARP, becomes rather despicable.

Post-processing can cause it (most TVs should have a "game mode" that disables it), but the pixels in an LCD panel also don't respond immediately. The 2-4ms response time you see on monitors with a TN panel is the gray-to-gray time. That's just the nature of the beast.

Game mode reduces it, but doesn't completely disable.

[PROMETHEUS]

If I had the option between playing a game on a PCB with 4 frames of lag, or on a custom build of MAME that emulated it but with only 1 frame of lag, I would choose the MAME version that responds in 1 frame.

I understand the principle of it all, but I'm going where there's less input lag, even if it's different from the original.

+1

[PROMETHEUS]

It's nothing like bashing a game because of the character art. It's like bashing a game because it looks so lazy and broken and rushed that they couldn't bother to have their sprites and backgrounds in sync. It's inexcusable.

The point was simple, people can use this build, but it should be invalid for competitive play, it isn't accurate, it isn't correct, and it looks really amateurish on YouTube vids.

Nah nah your exaggeration is huge here. The graphics glitch is incredibly minor. Have you tried the games ? I didn't notice anything in Ketsui, in DOJ the glitch is barely noticeable on a few select details of a few backgrounds. Its the kind of glitch you won't even notice on a youtube video. It's extremely minor, much more so than bad character art which is a big deal artistic-wise : I certainly prefer looking at DOJ's extremely slightly imperfect background on MAME than at MMP's sprites and levels on an Egret 3 haha.

[moozooh]

Have you tried the games ?

See Prome, Haze is a member of the team that makes MAME, but he doesn't play these games the way and for the same purpose you do, not even the way I do. In fact, he might not even play them at all, which is made obvious by his nitpicking at the graphical artifacts in the background which I have troubles noticing when dodging bullets coming from ten different directions. You have to be a dang pro to be able to fix your eyes on a wobbling sprite for enough time to become upset with that and not dying all over the place.

I mean, he's entitled to his opinion even when he doesn't know what he's talking about, but when he acts like it's his duty to convince MARP admins this version is unacceptable it becomes truly aggravating. :|

[Heartwork]

IseeThings' posts remind me of that Louis C.K. skit where he talks about how "everything is amazing and nobody's happy."

You're playing games that used to require a fridge-sized piece of hardware to run, and you're playing them for free, on your computer, at home, on the fucking internet. ShmupMAME is absolutely incredible, has made so many shmup fans' lives so much easier, and is developed by people who don't get paid a cent for their efforts on this project. Stop your damn complaining and enjoy the fact that people care enough about this community to provide things like this for free.

ps: huge thanks to Nimitz and whoever else has contributed to shmupMAME. You guys are champs.

[Drachenherz]

After reading through the last umpteen pages of meaningless bickering, I have come to the following conclusion:

Iseethings may have some valid points (i.e. the "original" MAME-build is not broken, Lag is sometimes in the original hardware etc.), but he should lighten up a little. He has made his point, several times, and he should now accept that other people don't share his opinion about what mame-version is the best for them. Seriously, STUFF IT! Repeating yourself ad infinitum doesn't make others take over your opinion. Deal with it and stop pestering this thread anymore, I think about 6 pages of bickering are more than enough. And believe me... this bickering doesn't validate the points you made anymore. In the beginning, yes, I thought what you've been writing made quite some sense. But now... It only makes you look like a selfrighteous and condescending prick.

[BPzeBanshee]

On the subject of LCD screens:

When I go to play street fighter 4 at friend's houses or elsewhere the lag is fucking unbearable at first. I get used to it tho. I feel worse lag in some instance offline on LCDs than when I play online at home, which should tell you how bad some screens are. Of course they now manufacture screens with very short responses, but I don't know how upscaling plays into that.

I can't stand playing Guitar Hero on my girlfriend's LCD screen for the same reason. Fortunately Guitar Hero has some strange Calibrate Lag function that solves the issue if tuned properly, but you dont have such a function with MAME and other games/programs because its a different beast.
I dont know precisely about Wireless controllers but I'd be willing to guess that's probably about 1 to 2 frames. The way they're made I dont believe battery power affects it since if its insufficient power to maintain optimum performance it just bummers out. Distance would be a big deal but in front of a PC I doubt there'd be any more than 1 frame or 1 frame every 5 seconds or something.

Really its things like these that should be debated and put to the test by the techheads here instead of having people like me do guesswork, but no, we get "This MAME is broken because it removes the framebuffer on certain games just like it says in the Readme" shit instead.

Speaking of which:

After reading through the last umpteen pages of meaningless bickering, I have come to the following conclusion:

Iseethings may have some valid points (i.e. the "original" MAME-build is not broken, Lag is sometimes in the original hardware etc.), but he should lighten up a little. He has made his point, several times, and he should now accept that other people don't share his opinion about what mame-version is the best for them. Seriously, STUFF IT! Repeating yourself ad infinitum doesn't make others take over your opinion. Deal with it and stop pestering this thread anymore, I think about 6 pages of bickering are more than enough. And believe me... this bickering doesn't validate the points you made anymore. In the beginning, yes, I thought what you've been writing made quite some sense. But now... It only makes you look like a selfrighteous and condescending prick.

I'm with you here Mr. Panda.
I've held my tongue since I last spat it in this thread to see how things go and it's just degenerated into an apparently up-himself MAMEdev slagging off a build and being a hypocrite with MARP in the process. Now I'm all for emulation accuracy and validity and all but its been plain obvious from the beginning what this build of MAME does to make less frame delay and really things have just turned into more repetitive bitching to see who's dick's bigger/who comes out on top/who's right and who's wrong/what looks better In My Trollin' Opinion.

We get the point already IseeThings, now stop repeating yourself and spend your time on something more valuable than bitching!

[Treasurance]

[StarCreator]

Call me thick or a tard but how can LCD screens give more lag compared to CRT? Is it all interpollation or is there lag even at native res?

Post-processing can cause it (most TVs should have a "game mode" that disables it), but the pixels in an LCD panel also don't respond immediately. The 2-4ms response time you see on monitors with a TN panel is the gray-to-gray time. That's just the nature of the beast.

I always thought it was more the fact that most scalers driving LCDs wait for a complete picture, does whatever processing needs to be done, then change the pixels on the screen all at once, rather than a CRT drawing it line by line as it receives it from the input device. The fastest scaler in the world can't make up for the delay caused by waiting for a complete frame, so there will always be at least single frame of lag if you're using a not-CRT.

My understanding is that hardware pixel response times at this point are so fast that they're not really an appreciable factor in the lag anymore.

[Despatche]

Going back to 0.99 (which somehow breaks!) and backporting the changes (which break, apparently!) is still a horrible idea (and for the millionth time I'm not using it anymore).

[Barrakketh]

I always thought it was more the fact that most scalers driving LCDs wait for a complete picture, does whatever processing needs to be done, then change the pixels on the screen all at once, rather than a CRT drawing it line by line as it receives it from the input device. The fastest scaler in the world can't make up for the delay caused by waiting for a complete frame, so there will always be at least single frame of lag if you're using a not-CRT.

If it's waiting for a "complete frame" at the screen's native resolution then that isn't scaling, that is deinterlacing. If you're using a digital input to drive the display (DVI, HDMI) then that shouldn't be an issue. If you're using an analog input like VGA or component cables (for instance, you're using an older Xbox 360 that lacks a HDMI port) then you're going to have that delay. And if you're using an analog input with a 1080p display you may very well get some scaler lag because I don't think component cables have the bandwidth for a 1080p60 signal.

And I suppose that particular aspect would be more properly called LCD input lag.

My understanding is that hardware pixel response times at this point are so fast that they're not really an appreciable factor in the lag anymore.

If people are excited about one less frame of lag from using Shmupmame, that same audience is going to consider pixel response times to be an appreciable factor.

Aside from that, pixel response times vary by panel type.

[austere]

If it's waiting for a "complete frame" at the screen's native resolution then that isn't scaling, that is deinterlacing.

No, not necessarily. Like you mentioned before, there's post-processing even at the native resolution. It is used to overcome the non-linear/time-invariant response of the polariser et cetera. This Samsung 275T is particularly bad in that regard, 2-3 frames (EDIT: perhaps >3) or so for 1920x1200 (native).

I don't think component cables have the bandwidth for a 1080p60 signal.

Depends on the length and quality of the coax cables. Even then, the system is limited by the ADC on the receiver rather than the cable itself, most of the time. It'll have to be a multiple of 124MHz@~12-bit, or less if they use phase locking. That's why most TVs don't bother, sadly, since component is superior to HDMI in certain situations but inferior for most people since it's 5 cables they have to connect rather than one.

If people are excited about one less frame of lag from using Shmupmame, that same audience is going to consider pixel response times to be an appreciable factor.

Given the input lag is usually an order of magnitude higher, I don't think this is an issue. Even then, a few ms will be impossible to notice. A single frame, yes, it's very noticeable. Just try and sense how much shorter a time you have to bomb, even if you correct for it most of the time the situation is hard to anticipate. I can't believe some people are doubting the ability for a human being to sense the difference. Even that Classic Game Room guy was able to sense a difference between Sonic Adventure on Dreamcast (3 frames of lag from the console alone) and on the xbox 360 (at most 2 frames input lag on the whole system), by noting that the controls are more responsive in the latter.

[friedrich]

Is anyone using a Wii controller (with GlovePIE) with this build?

I can't get it to work. Any help is appreciated.

[angrycoder]

Getting a crash on the new 3.0b version. Windows 7 64bit. Tried running as admin and using various compatability settings, no dice. Here is the crash info:


Problem signature:
Problem Event Name: APPCRASH
Application Name: mame32p.exe
Application Version: 0.96.0.0
Application Timestamp: 4d3c856d
Fault Module Name: mameplib.dll
Fault Module Version: 0.0.0.0
Fault Module Timestamp: 4d3c856a
Exception Code: c0000005
Exception Offset: 01940fa0
OS Version: 6.1.7600.2.0.0.256.4
Locale ID: 1033
Additional Information 1: 4c0d
Additional Information 2: 4c0d4d78887f76d971d5d00f1f20a433
Additional Information 3: 4c0d
Additional Information 4: 4c0d4d78887f76d971d5d00f1f20a433


Anyone else having a problem with this version?

[hipgnosis]

I have the same problem on my Windows 7 64 bit. Can't launch it at all.

[joeboto]

ive been getting that error 0xc00005 ever since i installed itunes.

[=SNC=]

If I had the option between playing a game on a PCB with 4 frames of lag, or on a custom build of MAME that emulated it but with only 1 frame of lag, I would choose the MAME version that responds in 1 frame.

I understand the principle of it all, but I'm going where there's less input lag, even if it's different from the original.

+1

If I had the option between playing dodonpachi on a PCB with 0 starting score, or on a custom build of MAME that emulated it but with 5.000.000 starting point, I would choose the MAME version that gives you 5.000.000 starting point.










No w8 it doesn't make sense.

[Lance Boyle]

http://en.wikipedia.org/wiki/Straw_man

[Udderdude]

Feel free to start your own Serious Business(tm) highscore thread, where only proof of playing on the PCB (which must be visible with proof that it's on even if running on a cab, can't have those pesky MAME PCs hiding inside one) is valid.

itunes.

Thur's yur prublem right thur.

[emphatic]

Hey, nimitz, any chance you will post a source update (v3.0c) on your blog for now and the binary later? Pretty please, etc.

[t0yrobo]

Getting a crash on the new 3.0b version. Windows 7 64bit. Tried running as admin and using various compatability settings, no dice. Here is the crash info:


Problem signature:
Problem Event Name: APPCRASH
Application Name: mame32p.exe
Application Version: 0.96.0.0
Application Timestamp: 4d3c856d
Fault Module Name: mameplib.dll
Fault Module Version: 0.0.0.0
Fault Module Timestamp: 4d3c856a
Exception Code: c0000005
Exception Offset: 01940fa0
OS Version: 6.1.7600.2.0.0.256.4
Locale ID: 1033
Additional Information 1: 4c0d
Additional Information 2: 4c0d4d78887f76d971d5d00f1f20a433
Additional Information 3: 4c0d
Additional Information 4: 4c0d4d78887f76d971d5d00f1f20a433


Anyone else having a problem with this version?

Same thing here, no luck with compatibility settings either.

[Iori Branford]

Here's a new issue, that doesn't exist in current MAME.

When I enable Sync to Monitor Refresh, I get unpredictable speed spikes where the game state jumps forward some frames, but without any break in the audio. It appears when game and audio desync, it's the game that plays catch-up with the audio -- if this can't be fixed, at least the other way around wouldn't so adversely affect gameplay.

[TransatlanticFoe]

Getting a crash on the new 3.0b version. Windows 7 64bit. Tried running as admin and using various compatability settings, no dice. Here is the crash info:


Problem signature:
Problem Event Name: APPCRASH
Application Name: mame32p.exe
Application Version: 0.96.0.0
Application Timestamp: 4d3c856d
Fault Module Name: mameplib.dll
Fault Module Version: 0.0.0.0
Fault Module Timestamp: 4d3c856a
Exception Code: c0000005
Exception Offset: 01940fa0
OS Version: 6.1.7600.2.0.0.256.4
Locale ID: 1033
Additional Information 1: 4c0d
Additional Information 2: 4c0d4d78887f76d971d5d00f1f20a433
Additional Information 3: 4c0d
Additional Information 4: 4c0d4d78887f76d971d5d00f1f20a433


Anyone else having a problem with this version?

I'm running Win 7 64 bit as an Administrator user, installed in a subfolder of C:\ (not Program Files or Program Files (x86)) and it works fine, so the OS isn't the problem.

UAC might be mucking around if you've got it turned on? Also check mameplib.dll is in the same directory as the executable.

[bcass]

Hiscore.dat has now been updated to support these new titles:

http://highscore.mameworld.info/download.htm

[SPYvsSPY]

Here's a new issue, that doesn't exist in current MAME.

When I enable Sync to Monitor Refresh, I get unpredictable speed spikes where the game state jumps forward some frames, but without any break in the audio. It appears when game and audio desync, it's the game that plays catch-up with the audio -- if this can't be fixed, at least the other way around wouldn't so adversely affect gameplay.

Looks like they've also removed some sync options in mameFX also? could be more than a Shmupmame issue? http://mame32fx.altervista.org/home.htm

[captpain]

Here's a new issue, that doesn't exist in current MAME.

When I enable Sync to Monitor Refresh, I get unpredictable speed spikes where the game state jumps forward some frames, but without any break in the audio. It appears when game and audio desync, it's the game that plays catch-up with the audio -- if this can't be fixed, at least the other way around wouldn't so adversely affect gameplay.

You're worried about a potentially lag-adding option adding some weirdness to a version of MAME that only exists to remove lag?

Erm... ok, well use a different version of MAME.