Article: What is ShmupArch? Why Does it Matter?

[Mark_MSX]

Hi everyone!

So, to help clear up some questions about ShmupArch and maybe help answer some questions you all may have, I have written up this article.

http://electricunderground.io/what-is-s ... it-matter/

Enjoy!

[WelshMegalodon]

Many months ago, Xyga lamented the absence of a beginner-friendly guide to reducing lag on an LCD display with GroovyMAME. Now, in the continued absence of such a guide, a goofy savestate-abusing exploit implemented in a trainwreck of an emulator frontend with one of the worst GUIs I've ever had the misfortune of using has somehow become the "best" low-lag solution for emulation. How about that?

[ATTRACTS]

Many months ago, Xyga lamented the absence of a beginner-friendly guide to reducing lag on an LCD display with GroovyMAME. Now, in the continued absence of such a guide, a goofy savestate-abusing exploit implemented in a trainwreck of an emulator frontend with one of the worst GUIs I've ever had the pleasure of using has somehow become the "best" low-lag solution for emulation. How about that?

Needs more salt.

[Mark_MSX]

Many months ago, Xyga lamented the absence of a beginner-friendly guide to reducing lag on an LCD display with GroovyMAME. Now, in the continued absence of such a guide, a goofy savestate-abusing exploit implemented in a trainwreck of an emulator frontend with one of the worst GUIs I've ever had the pleasure of using has somehow become the "best" low-lag solution for emulation. How about that?

There's no denying that the UI in Retorarch is a hot mess, it definitely needs improvement for sure. I've put together a guide that helps with the setup, but I agree that it is unnecessarily complicated. As far as GroovyMAME is concerned, once I get the proper graphics card and a spare CPU, I am actually going to put together some content on GroovyMAME and do some testing. I haven't done a lot, but I have messed around with GrooveyMAME on LCD and found that it is about as responsive as shmupmame, but it can't get down to as low as 1 frame. I'm not exactly sure why you'd characterize Run-ahead as a "goofy savestate-abusing exploit" but right now it appears to be the strongest method of removing latency from shmup emulation. The feature is certainly unrefined right now, but I expect that it will continue to improve along with RetroArch.

[Xyga]

Many months ago

Many months later, Xyga doesn't give a crap. Things like shmupmame and now this, is all this fallen forum deserves.

[Mark_MSX]

Many months ago

Many months later, Xyga doesn't give a crap. Things like shmupmame and now this, is all this fallen forum deserves.

Personally, I see nothing wrong with the community having more emulation options available, especially in the area of lag reduction. I don't really understand the negativity towards shmupmame or shmuparch. Many great community records were set using Shmupmame. In DDP for example, Prometheus used Shmupmame. And I do know that, after trying it, Jaimers now uses Shmuparch.

[donluca]

The issue is that shmupmame is fundamentally broken due to both using a very old MAME version and a series of hacks to get the input lag as low as possible.

Shmuparch on the other hand may be even worse as it uses FBA which is known for going for performance over accuracy by using all kind of shortcuts in order to make the games work on most devices possible.

What could actually be interesting is packaging GroovyMAME as a core for retroarch and doing all the configuration via retroarch so that a release like shmuparch could be done.

GroovyMAME with a beefy CPU is still the way to go for those who are serious enough to invest some time into researching GM and how it works, especially with a CRT.

For all the others, shmuparch is actually a great alternative. At least it's based on a modern FBA core rather than an old, broken MAME version.

[Mark_MSX]

The issue is that shmupmame is fundamentally broken due to both using a very old MAME version and a series of hacks to get the input lag as low as possible.

Shmuparch on the other hand may be even worse as it uses FBA which is known for going for performance over accuracy by using all kind of shortcuts in order to make the games work on most devices possible.

What could actually be interesting is packaging GroovyMAME as a core for retroarch and doing all the configuration via retroarch so that a release like shmuparch could be done.

GroovyMAME with a beefy CPU is still the way to go for those who are serious enough to invest some time into researching GM and how it works, especially with a CRT.

For all the others, shmuparch is actually a great alternative. At least it's based on a modern FBA core rather than an old, broken MAME version.

I can understand that. Ya, no doubt FBA has its limitations for sure. I can understand the concern. The good news is that there is a MAME core for Retroarch. Unfortunately, it does not support run-ahead right now and is way too laggy to recommend. However, there is progress being made in that department. I completely agree that, once that core gets run-ahead support, it will be the go to solution over FBA for sure. I'm actually really excited for that to happen.

Also, if you do not want to use FBA, you can actually use the MAME 2003 core, as it supports run ahead. Of the two, I have found that FBA performs better generally, but Mame 2003 core is another option with ShmupArch.

[zakk]

If your concern is accuracy, mame 2003 is no better (and probably worse) than FBA.

If your concern is speed you should probably just stick to FBA.

[donluca]

This is just my own *very* personal opinion, but the day all those old MAME versions get nuked I'll probably be celebrating somewhere.

The only reason those old, buggy, crappy, hacked piece of crap exist is to give users an alternative to FBA when you have very limited computing power, such as RaspberryPi and the likes.

The sooner we get rid of those, the better. FBA has made huge improvements not just on the performance side, but also on accuracy and it runs better than any old, hacked MAME version.

Sweet Lord, was one of those old versions based on 0.37b? That was like... uh... 15 years ago I think.

Anyway, rant over, back on topic.

[Mark_MSX]

Yes I agree. I strongly prefer FBA over these old MAME versions, the performance just seems much for reliable. I definitely recommend using the FBA core for now.

[Xyga]

@donluca it's worse than just old versions and hacks, the premise (all games down to 1 frame) is absurd to begin and shows that even at 20 years of emulation people still don't get shit and especially not MAME, would they be complete beginners or gaming e-celebrities. I had written a detailed reply and then recalled that I've wasted my time on topics like that, too many times for fucking nothing. So *fart*

[Mark_MSX]

My question is who are the gaming e-celebrities playing shmups?

[cools]

http://electricunderground.io/what-is-[u][i][b]shump[/b][/i][/u]arch-why-does-it-matter/

Fussy about the details GroovyMAME elitists represent.

[Keade]

Many great community records were set using Shmupmame. In DDP for example, Prometheus used Shmupmame. And I do know that, after trying it, Jaimers now uses Shmuparch.

The fact that they are hacks that give a big advantage make these "solutions" even more misfit for "serious" competition than "regular" emulators, imho.

My question is who are the gaming e-celebrities playing shmups?

Prometheus ?

[poptart]

Prometheus ?

What? In what reality is he an e-celeb? matter of fact i can't think of anyone in the entire genre that'd qualify for such a title.

The fact that they are hacks that give a big advantage

Your scores will still be shite lad so don't worry about it.

@donluca it's worse than just old versions and hacks, the premise (all games down to 1 frame) is absurd to begin and shows that even at 20 years of emulation people still don't get shit and especially not MAME, would they be complete beginners or gaming e-celebrities. I had written a detailed reply and then recalled that I've wasted my time on topics like that, too many times for fucking nothing. So *fart*

How does any of this affect you? 1. just play your beloved groovymame and 2. you're more the collecting type so don't sweat it.

[Xyga]

Who's that cunt ? oh yeah probably coming from that cesspool. So they're still polluting here lol.

[Mark_MSX]

Many great community records were set using Shmupmame. In DDP for example, Prometheus used Shmupmame. And I do know that, after trying it, Jaimers now uses Shmuparch.

The fact that they are hacks that give a big advantage make these "solutions" even more misfit for "serious" competition than "regular" emulators, imho.

My question is who are the gaming e-celebrities playing shmups?

Prometheus ?

Big advantage over who exactly? Japanese PCB Players? Oh I forgot about how we've been completely destroying Japanese scores since Shmupmame was released. Or are you talking about people using grooveymame? Are you guys the real "serious" competitors, is that it? Let me just get this straight for future reference. So players like Prometheus, Jaimers, and Bananamatic who play these games and set high scores on this forum, they're just hacking cheating "e-celebrities," but you guys using grooveymame are the real deal? Good to know.

Glad I've been educated. I'll be sure to let them know that they better remove their scores.

I'm really happy to hear we're cutting down our accepted scores to people only using grooveymame, because our scoreboards are just bursting with activity as it is.

Really though, none of this talk sounds constructive or genuine. It's all coming across to me as you guys taking shots and trying to show me how smart and cool you are. I have an interesting checking out grooveymame and seeing what I think. But now that I know it is the ultimate perfect solution, I just feel like a fool. And here I was thinking the shmup emulation is still an evolving process, where it is worthwhile to explore new options and see what works. I mean, could it be remotely possible that maybe, just maybe, added latency is an unnecessary part of the emulation process? NO.

Lag is good. Lag is beautiful. I have seen the light.

[Feedbacker]

Shmupmame users have a huge advantage over groovymame users because they can use save states without having the program randomly crashing after few minutes of use.

[poptart]

Who's that cunt ?

Listen i'm sad to that your failing eyes won't be able to see your collection in a few years time but don't be down, maybe you can pioneer the braille spinecard for your sick grabs!

[Xyga]

[donluca]

Just out of curiosity... are leaderboards with highscores from shmupmame the same as people with PCBs? Just asking.

I'm somewhat tempted to ask what about recent CAVE shooters where even the Xbox360 ports aren't completely accurate, but I guess I'll keep that for another time.

I mean, anyone can use whatever the hell they like. As far as high scores go, imho the important thing is that everyone uses the same platform, be it shmupmame, shmuparch or GM.

There are several communities focused on single game scoring which are *very* strict about what platform/software should be used to achieve a submittable high score. I'd be surprised if the shmups community would be more "relaxed" in those terms.

[Keade]

Lag is good. Lag is beautiful. I have seen the light.

Of course lag is hideous. You are talking to someone who have waited for ages to eventually switch directly from a PC CRT to a 120/144Hz LCD, and who cringes at 30fps or software frame caps
However, most shmups on MAME have only 2 frames of lag (according to that thread) though, which is usually fine unless your setup sucks. Dodonpachi, which is one of the game you showed, is an extremely responsive game.
Weren't there a few games (okay maybe just a few...) where runahead could eventually have lower total lag than the real thing (PCB/CRT) ? This is where it is most similar to cheating, imho, which is way I said it could be "misfit" for scoring. tl;dr: I think the input lag issue is usually exagerrated (not always: Mars Matrix is horrid) and hacks are not an actual solution.

Anyway, as much as I stay away from Retroarch and think the input lag issue is exagerrated, the article was a good read.
Support for runahead in RA with recent MAME releases/core will not become widely available imho, because as you probably know, runahead requires you to be able to run the game *2, *3 speed and so on. This is possible for a few games, but for others like Mars Matrix, MAME will run at about 250% speed on a decent rig, not more. The dirty way to make things run faster is to use decades old releases, but then you are compromising more on accuracy/features.

[zakk]

The shmups community is absolutely more relaxed in that regard. There may be pockets of people that are more strict, but they are in a vast minority. Just go peek at all he HS threads here, you're not going to find many that keep separate tables for PCB and 3 versions of mame or whatever. It's too much work and who cares when most of the scores are like 50% or less of a 'real' WR. It's more common to split tables between "arcade" (including mame and all variants) and home console/pc ports if they're known to have inaccuracies/scoring differences.

Ain't no one screaming about someone using shmupmame and beating their rank #35 score because of it being 'unfair'.

If someone is genuinely approaching WR territory they may seek out a real PCB just to be 'legit', but that's not most people.

[Shepardus]

I doubt that input lag has much of an impact on scoring at all; what matters more is that you play on a consistent setup and get used to it rather than having the absolute lowest input lag possible. A couple frames is not a lot of difference, shmups don't typically demand frame-perfect reactions anyway. It can affect your enjoyment of a game if the controls feel sluggish or unresponsive, though, so I would recommend lowering input lag to the point where it feels comfortable to play. To me, vanilla MAME has offered good enough results for pretty much all games besides Mars Matrix and Guardian Force.

[Jaimers]

The fact that they are hacks that give a big advantage make these "solutions" even more misfit for "serious" competition than "regular" emulators, imho.

If anything, this brings PC closer to PCB. Since PC has more lag by default than a PCB hooked up to a cab with a CRT monitor.

A couple frames is not a lot of difference

You'd think that, but it really makes a huge difference. After playing DDP in Retroarch a bunch, I tried it in Shmupmame 4.2 the other day and it felt like almost unplayable. That's how big of a difference it was.

[colour_thief]

I don't understand who these lag apologists are. I own plenty of cabs and PCBs, but if there is a way of playing without lag in a game I am competitive in sign me the shmup up! I don't even care about being on the same leaderboard or have any desire to argue the score is comparable.

Where were you guys when M2 ported Futari and removed lag? Ain't nobody turning the lag option on. lol

Next you'll argue it doesn't count unless your buttons are sticky and your stick has cigarette burns...

I do wish something better than retroarch supported this though.

[moh]

I don't understand who these lag apologists are. I own plenty of cabs and PCBs, but if there is a way of playing without lag in a game I am competitive in sign me the shmup up! I don't even care about being on the same leaderboard or have any desire to argue the score is comparable.

Where were you guys when M2 ported Futari and removed lag? Ain't nobody turning the lag option on. lol

Next you'll argue it doesn't count unless your buttons are sticky and your stick has cigarette burns...

I do wish something better than retroarch supported this though.

[donluca]

Where were you guys when M2 ported Futari and removed lag? Ain't nobody turning the lag option on. lol

I do

[cools]

Emulation gives me a way of enjoying expensive, fragile games where I can't afford the risk of them breaking.

Meaning accuracy matters, including lag. I have no interest in reducing lag beyond what the original hardware provided.

From the article:

Run-ahead takes a game like Battle Garegga, which previously had 4 frames of lag in emulation, and chops it down to 1. BEAUTIFUL! Now we’re talking!

Good emulation adds no lag to Battle Garegga, it's inherent on the original hardware. Same for Batrider and Bakraid. The lag in these games hasn't affected their quality or reputation in the 20 years since they were released - millions of people have enjoyed them with the lag present.

If Raizing themselves were to release a patch for the PCB that dropped the lag to 1, I'd wholeheartedly embrace it as a "final version" of the game. Run-ahead is a bastardization and wrecks the experience.