Retroarch Accuracy (Slowdown)

[Mark_MSX]

Happy to help my dude Here are the answers you seek:

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

[casualcoder]

That's great, man! It'll read it with my morning coffee tomorrow.

Just wanted to note I was testing out Guwange again and confirmed runahead should be set to "3."

I've tried to calculate runahead properly with many games but noticed some actions take longer to come out than others.

So I tend to go with the quickest action and use that as the baseline to calculate runahead.

In the case of Guwange, the shot takes 4-5 frames to come out, but movement takes 3.


I usually go by character movement anyhow as that's the one that would mess with the experience the most if the runahead isn't set right.

And also, it's where lag is most detectable.

[Xyga]

To be honest I am not sure how RetroArch is able to run these games at non-60hz frame rates as smoothly as it does. It is interesting and kind of mysterious. I do disable vsync when I play and there are some games, like garegga, that have some screen tearing from time to time. But with DDP, there is no screen tearing at all, or none that I have noticed (and I've played DDP on RA alot at this point).

Here's a quick snippet I just grabbed from my stream. Sorry for the slight pixelation and occasional stuttering, that's my stream not the game.

https://www.twitch.tv/videos/380358625

Nothing mysterious, you disable vsync, that's why. Which means it's not complete, some games will show tearing and some less depending on one's setup tolerance, and where people don't see it they will call it smooth (individual honesty in this is what it's worth, shmupmame sync-less users pretending they don't see tearing among other issues always made me laugh). If someone's honestly not experiencing any tearing nor judder at all with an off-60Hz game while he has purposedly disabled vsync or syncrefresh, then there might be something somewhere forcing sync/buffer anyway, which could be another emulator core setting, or RA setting, or OS related even.
There's a reason why vsync exists; because that kind of source material with odd refreshes can't fully keep its picture together by itself on today's fixed 60Hz displays without it period.
Keeping it on without lag penalty is possible with GM and RA and a number of other emus, rather with what's called syncrefresh in mame's terms or equivalent (vsync at the game's rate rather than the system's/display's) but only under the condition it's run on an able setup; crt/free-gsync/non-edid-locked-lcd, and that works without much if any lag penalty at all despite what some stubborn people still believe.

Additional edit: this is only speculation but one thing that might give credit to at least an attenuation of tearing while any form of vsync/buffer is off, would maybe be related to run-ahead. I've been wondering if it is not somehow acting more or less like a buffer for sync while performing its frames temporal trick.

So that's it, you can't have it all in arcades emulation without that, whether you use RA, GM, shmuparch, shmupmame, whatever, you'll have to sacrifice either correct refresh speed, or stable synced picture, or drivers proper code.

RA/shmuparch already managed to convince a number of people that it doesn't matter that the original game's legit lag is eliminated, regardless of a certain consequences, I'm not here to debate that again, but to say it'd be good not to make people misunderstand another thing on top of that, or you'd further sound just like another snake oil salesman, you sure don't want that, do you?



About the video: it is kind of bad quality and 576p@50Hz (!) I can't see clearly enough what's going on as it drops part of the video information.

[CloudyMusic]

Nothing mysterious, you disable vsync, that's why. Which means it's not complete, some games will show tearing and some less depending on one's setup tolerance, and where people don't see it they will call it smooth (individual honesty in this is what it's worth, shmupmame sync-less users pretending they don't see tearing among other issues always made me laugh).

In my limited experience, 120Hz monitor refresh does wonders for this. The only games where I've had screen tearing pose an actual gameplay hindrance are Psikyo games due to the bullet speed, but switching my monitor to 120Hz with v-sync off basically made the tearing imperceptible.

The tearing created by v-sync off has never even come close to being an issue for me in slower games, even at 60Hz.

[Xyga]

Yes doubling the rates over a 120+Hz display is a known way to mitigate the tearing of non-synced video.

The tearing created by v-sync off has never even come close to being an issue for me in slower games, even at 60Hz.

Not for you that's good, personally I'm among those who can't stand tearing even a little. In any case displaying video with tearing is not the normal way and not what we should expect, do arcade games have tearing in their original normal environment? no. Anyway we have means to completely eliminate tearing today even without suffering more lag.

What matters is to be clear and not pretend things that aren't true, RA like most emu solutions cannot perform lag reduction and accurate smooth scrollings at proper real rates at the same time using a setup/display that technically can't, implying it does would be lying or at least deliberately keep silent about the fact and the downsides of making a choice of either playing with tearing or eliminating it by use of sync.
Again all arcade emus give you a choice on the matter, and neither is without its downsides as long as no setup that can display clean non-60Hz rates is used, something is sacrificed, there is no magic that these emus can perform by themselves (so far...this could change if beam racing aka frame slice ever becomes a practical reality)

[casualcoder]

@Mark_MSX read the article, I get it now... and I think it's great!

Honestly reading it was nice because it confirms the same experiences I've been having and I think it's a great thing to catalogue and "curate" the problems/solutions, particularly from a shoot-em-up perspective.

I too think the lag issue has been a big burden on Mame setups and wasn't sure why the lack of enthusiasm and support of RA and runahead.

And the confusion surrounding RA certainly doesn't make things easier.


I get it's not an actual development fork, but I think it's a great idea.

Is there a place/post where people can contribute their findings/help maintain and update games/configs?

Another question, how do you plan on merging base software updates with "ShmupArch"? I imagine it can be quite task ensuring things stay up to date and yet don't break any previous settings, especially since none of this is source controlled.

[Xyga]

I too think the lag issue has been a big burden on Mame setups and wasn't sure why the lack of enthusiasm and support of RA and runahead.

Yes the problem is there. Mark's goal is clearly to make people confused about the lag question, there's more than one source of lag in emulation, the driver's proper emulating the game and when accurate matches that of it running normally on its original system/pcb, that added by video rendering lag depending on the backend used and sync settings, and that related to the inputs layer where applying that runs along the video thread.
Seeking to reduce lag for accuracy means trying to eliminate the additional delay created by the latter two on top of the game's normal delay.
Enter run-ahead, a clever lag reduction technique, which allows not only to tend to the legitimately unwanted delay, but to break past and reduce the game's normal delay as well.
The issue is that if you use run-ahead like this, not only it means an advantage over people playing in real conditions (benefit of better reactivity than the game is supposed to produce), but also it does not appear in videos/streams, besides the question of emulation accuracy both situations legitimately raise questions about fairness. In the first case this can eventually allow some to compensate for a laggy display, in the second it's rather hard to justify since it's invisible if not signaled at the beginning of a run.
So Mark with that pre-configured RA build, since it's nothing else despite having a brand name slapped on it, decided with a number of people that this build configured to floor delay as low as possible with what RA's run-ahead allows, wasn't a problem, and that people who protest are somewhat illogical because of course who likes lag !?
Yes of course, in the position of the salesman if you deliberately avoid to mention what's wrong with your merchandise it's easy to bring people unaware of these 'details' to follow you and to make the ones who do care and raise concern to pass for negative bad guys.
Get some high-scorers with a name to put their seal of approval with a couple of tweets or replays to join the bandwagon and poof! the campaing promotion works, the whole thing is expected to become the new standard for emulation and players around the world, regardless of facts some annoying grumpy party-poopers bastards like me brandish.

But since you seem to have taken the bait anyway rest assured I'm not trying to convince you, casualcoder.

Sometimes I really dislike mamedev's stubbornness and tunnel vision on some topics, I think they've endangered the present and future of emulation their own way by being too deaf/blind, rigid and away from gamer's reality, too convinced that pure developer logic is the only one valid.
But all that RA achieved with run-ahead, the only thing they've really brought besides the great shaders support and that's not just yet another appropriated and rebranded work, is to further stain the emulation scene by giving unaware idiots who don't give a fuck about anything, the power to appropriate even a portion of the gaming scene as a consequence no matter if it's with a set of hazardous fucked up rules that shit on everything that was before. And knowing them and their logic if they can manage to further appropriate, wreck, and rewrite reality until their egos are satisfied, they fucking will. Snake oil is good, that's what we're supposed to accept these days.

Fingers crossed that beam racing / frame slice will really happen some day, or that variable refresh becomes the standard on pretty much every system globally, to finally end that joke.

Note: to try and be fair with RA devs, maybe that wasn't their original intent, but as they've witnessed the immediate consequences, they did nothing to correct or mitigate either, so it's impossible not to blame them.

[casualcoder]

Yes the problem is there. Mark's goal is clearly to make people confused about the lag question,

I mean, I've been away from these parts for a while so I'm not going to pretend I know what's up.

But, I'm not getting that impression from Mark's efforts at all.

It would seem an insane proposition to consciously try to lead people astray.

And on top of that, it's a whole lot of thankless work... even if (in your judgement, it seems) provides no value.


I try to assume the best in people first until evidence to the contrary.

If anyone were trying to literally sell this as a separate piece of software, then yeah that would draw my ire as well.

I don't see that being the case here.


In terms of the question of lag reduction and competition...

My opinion is fairly blunt on this as well. This whole business, mame, shoot-em-ups, high scores. It's a niche. A niche within a niche.

Competitive gaming in general is gaining momentum, but the issues surrounding legitimacy of score keeping is something that corner has always had to deal with.

I see these latest developments (why stop here, let's also talk about digital video editing as well) are just the latest in a long history that will always be tarnished with legitimacy questions unless competition is standardized.

From my point of view, I only compete with myself, really.

Yes, I've posted a few high scores on here and so forth. Mainly it's about community. It's nice knowing there are others attempting these oddball tests of skills around the same time as yourself.

It can be nice as a motivator.


But in general, I just try to test my skills relative to what I think I'm capable of when unfettered with external factors.

Is this more like wearing shoes vs heavy boots when playing soccer? Or more like taking steroids when body building?

I don't know. I don't even know if it matters.

But I do know it prohibits me from enjoying the experience if I feel the experience could be made better.


As an interesting side note, I just took upon the task of defeating Mike Tyson on NES Punch Out yesterday.

I did it yesterday after wanting to do it for... at least 30 years.

At first I did it in RA with runahead set to 3 frames.


It's hard to say how many frames the game has of built-in latency, but I do know my display has 1 frame, and then there is at least another frame of controller/emulation latency on top.

I felt like maybe my "victory" was a bit illegitimate considering it's a very special case scenario. Almost as pure an example of the question of lag reduction fairness as you can generate.

So I tried again with 2 frames runahead and beat Tyson pretty much as easily. I figured I'm still within the 1 frame that probably would have been what you'd get on CRT/original hardware.


And, further to the point. I found that the 3 frame latency was a good psychological motivator, considering questions of latency didn't mire every attempt.

It kept me playing, and getting better.

Now I'm quite sure I can even do it without latency compensation. Though, it would probably take some more tries.

[Xyga]

'Sell' was employed figuratively. There is no purpose in ShmupArch beyond passing as a creation where it's nothing but a preset RA with abusive settings which run-ahead unfortunately allows, I don't see where's the 'work' here.
The work is rather there; the guy's set that thing up, put his brand name on it, made a website, discord, casts, used all means of social media and articles and people to promote the acquisition of his own share of the niche, shmuparch is only one of the means.
Would have he made that using a pre-set build that doesn't shit on emulation accuracy and fairness I would actually have had no criticism, in fact I think this would have been a good thing, that would have very well been possible but no, he deliberately chose 'eliminate all the lag!!!' regardeless of the consequences, and I don't believe for a second that he never though of what could be wrong with it, but since it's like honey obviously that was going to be popular.
Just reading his articles and how he presents shmuparch around to people clearly supports a deliberate narrative carrying an intent to bend people's opinion towards accepting a new set of rules packed together with the issues, whether they are aware of those or not, better if they aren't. I know a salesman speech when I see one.
In fact I don't care about his ambitions, he can become the internet media guru of all shmuppers on earth for all I care that wouldn't bother me in the least, but he uses a nasty tool, that 'shmuparch' as it is, to achieve them, and even though the original sin is RA developer's, this additional promotion of the wrong side of RA further pollutes emulation which is something that deserves more to advance, to progress and improve, not regress with patched up hack job builds. People who don't give a crap and don't see the problem supporting bullshit damaging emu solutions only contribute to sap the legit hard work of real developers like mamedev's or groovy's, and in fact RA would be much less of an issue if its developers had a minimum of awareness and ethics in them, it's bad for everyone that they don't.

[casualcoder]

'Sell' was employed figuratively. There is no purpose in ShmupArch beyond passing as a creation where it's nothing but a preset RA with abusive settings which run-ahead unfortunately allows, I don't see where's the 'work' here.
The work is rather there; the guy's set that thing up, put his brand name on it, made a website, discord, casts, used all means of social media and articles and people to promote the acquisition of his own share of the niche, shmuparch is only one of the means.
Would have he made that using a pre-set build that doesn't shit on emulation accuracy and fairness I would actually have had no criticism, in fact I think this would have been a good thing, that would have very well been possible but no, he deliberately chose 'eliminate all the lag!!!' regardeless of the consequences, and I don't believe for a second that he never though of what could be wrong with it, but since it's like honey obviously that was going to be popular.
Just reading his articles and how he presents shmuparch around to people clearly supports a deliberate narrative carrying an intent to bend people's opinion towards accepting a new set of rules packed together with the issues, whether they are aware of those or not, better if they aren't. I know a salesman speech when I see one.
In fact I don't care about his ambitions, he can become the internet media guru of all shmuppers on earth for all I care that wouldn't bother me in the least, but he uses a nasty tool, that 'shmuparch' as it is, to achieve them, and even though the original sin is RA developer's, this additional promotion of the wrong side of RA further pollutes emulation which is something that deserves more to advance, to progress and improve, not regress with patched up hack job builds. People who don't give a crap and don't see the problem supporting bullshit damaging emu solutions only contribute to sap the legit hard work of real developers like mamedev's or groovy's, and in fact RA would be much less of an issue if its developers had a minimum of awareness and ethics in them, it's bad for everyone that they don't.

Well, I feel like I get Mark's motives in the sense that it illustrated the very problems I came up against when trying to use Retroarch, and especially in the context of playing my beloved shoot-em-ups.

If I had seen Mark's efforts, it probably would have saved me a lot of time and personal testing.


I think it's totally fair to have your own opinions on emulation, lag reduction, competitive fairness, etc.

I also think it's worth recalling to yourself that these are just opinions.


The fact that people don't agree means we're all just trying to find the right course of action in all of this, bouncing ideas until people collectively judge and arrive at a set standard.

Maybe you guys have underlying personal animosity behind all of this, but on the surface, it just seems like the same types of people wanting the same things but going about it differently.


I guess I'm the perfect example of a wish-washy centrist as I see the value in both of your ways of working through this particular problem.

[Xyga]

If even hard technical facts are just a matter of 'opinion' then right, the Earth is flat, why not. And if being good at selling snake oil is 'effort' then fuck everything, I'm going to patch up a build too, call it 'MiracleMAME' polish my narrative and start my own internet media package, then why not raise some crowdfunding shit so sheep thankful for my 'work' can finance my life. Why not.

Again there are better, legit solutions, I've already made that clear in details. I've seen enough people in recent years who support hack shit job emus claim all of sort crap and support things like RA and shmuparch, but never care to actually use ther brains to learn how this whole deal works and actually look into the better available solutions themselves, either because they pretend it's too complicated or make excuses saying it doesn't work because they've failed trying, or they're so cheap that they don't want to spend money on a fitting setup. But as I said earlier in this thread, curiously they always find the energy to reject and argue against reason.

I'm not specifically saying this against you casualcoder, I have nothing against you, it's just that this has been a case for a number of years and it's so stupid that the wrong stuff always wins that it pisses me off, even though yeah, sure we're only talking about emulators, video games and niche at that. And people will do what they want anyway.

[DietSoap]

Mark's goal is clearly to make people confused about the lag question

Not sure this is completely accurate...

[el_rika]

Happy to help my dude Here are the answers you seek:

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

Hey man, could you perhaps help me out with 2 issues here?

I am on Android (good phone that can play cv1000 in Mame Retroarch full speed) and i was wondering if:

1. I can set some things in Retroarch to allow me smooth acrolling while keeping the original game's intended speed?

2. I can unlock (via cfg? or nvram?) game modes such as Ultra in Futari 1.0 rom and Strong Style in Daiffukatsu 1.0 / 1.5, which can originally be unlocked via special code (which i can not do as i don't have access to 2nd player).

Thank you!

[Shepardus]

2. I can unlock (via cfg? or nvram?) game modes such as Ultra in Futari 1.0 rom and Strong Style in Daiffukatsu 1.0 / 1.5, which can originally be unlocked via special code (which i can not do as i don't have access to 2nd player).

You could unlock it on a PC and transfer the nvram files over to your device.

[el_rika]

2. I can unlock (via cfg? or nvram?) game modes such as Ultra in Futari 1.0 rom and Strong Style in Daiffukatsu 1.0 / 1.5, which can originally be unlocked via special code (which i can not do as i don't have access to 2nd player).

You could unlock it on a PC and transfer the nvram files over to your device.

Oh, that's great news.

But what exactly should i do..? (step by step)