SHMUPMAME & Front-end

[BPzeBanshee]

I recall Garegga having a lot of lag, too, but just playing around with the Iron Mackerel in both Shmupmame 2.2 as well as MAME Plus! 0.146u5 (Aug 21 2012), there doesn't seem to be any difference, I can tap-swap direction quickly in either.

I didn't use frame skip anymore because that is a flawed metric for actual play.

2.2 did nothing to Garegga IMHO. While toaplan2.c had been modified the actual 'radical change that Nimitz was going to write a blog post about' only happened in v3.0, and Barrakketh's builds that claimed to have Lagless Garegga used 2.2's base and did nothing to reduce response time in my prior tests.

Here's a better example game to play: ESPRade. It's been 'lagless' since the original Lagless MAME release so no disparity between releases, and I swear to you its response times are improved compared to standard MAME: it was plain obvious to me from the first stage (and there's no real background-sprite stuff there so the desync is not even noticeable) and isn't difficult to spot in frame-by-frame advance either unlike Garegga which from the last discussion between you and me concluded was all over the place.

[Erppo]

I won't let people talk shit about one of the greatest voluntary work ever that propulsed the arcade genre in the west like nothing else has.

That was actually Touhou.

[Giest118]

So the only person here who has really huge problems with shmupmame is trap15, who is known for some nifty/awesome romhacks. His experience making those suggests that he has more intimate knowledge of the technical side of things than most people. Therefore his opinion is likely colored by that and doesn't have much to do with actually playing the games. Which is a fine opinion to have, don't get me wrong, but it seems to me that gameplay-oriented people will be worried more about things that actually affect gameplay, such as input lag. Shmupmame does pretty good in those regards, so for most people, shmupmame should be more than adequate.

Or I could be talking out of my ass. All I know for sure is that I use shmupmame and I love it. But that could be because I'm a faggot who only plays Cave and Psikyo games.

[Op Intensify]

I think the correct opinion is that Shmupmame is preferable for certain games with severe input lag problems (Psikyo, Seibu, Raizing from Garegga onwards) and less than desirable for others (Cave).

[PROMETHEUS]

It emulates the games very accurately, significantly closer than any console port.

Now, I understand your point about it being "accurate". You mean it plays with lag closer to the PCB. This, I can agree with you on, and is objectively true. However, console ports are either equal or better in this regard, since they neither have to hack around the lag of the host system (like you have to do for PC), and they don't need to emulate any sprite buffer since there isn't any (so they get even better reaction time than ShmupMAME, since you remove lag from the PC platform from the equation. It also doesn't have to cause background desync unlike ShmupMAME).

There is a barely noticeable graphics glitch as a tradeback (thin black line that may flicker a bit on some rare parts of backgrounds).

That's a very small issue that I don't think I've ever had; in fact, I think that's related to other hacks in ShmupMAME. The real issue with remove the sprite buffer is that it causes sprites that are "attached" to the background to move 1 frame delayed from the actual background, since the code would normally correct itself for the buffer. This is completely inaccurate; this never happens on a real PCB.

I'm always very sensitive with things like input lag, and the difference between my PC setup and a cab with PCB in it I can't even feel at all (I can feel the difference between 1 and 2 frames of input lag). I've heard this sprite buffer stuff before, but how much lag are you talking about ? Have you tested it ? Can you make a demonstrable comparison of PCB/MAME/Console Port ? From my own experience, I can bet that if increased input lag there is on PC, it is so small that you can pretty much call it insignificant.
Again, console ports have many "inaccuracies", be it with sound, graphics, or even gameplay mechanics and speed. Do we need to compare ports with MAME on those grounds again ? Do you not realize how much more speed inaccurracy affect playing the game than the most minor graphics glitch that every other guy even notices ?
(That said, I think many ports are good enough, the inaccuracies are minor enough. Which is why it is completely laughable in my view to call MAME inaccurate)

I won't let people talk shit about one of the greatest voluntary work ever that propulsed the arcade genre in the west like nothing else has.

[citation needed]
Seriously, where are you getting the idea that ShmupMAME "propulsed the arcade genre in the west"? It's as dead as it's ever been, and no shitty version of MAME is going to change that.

Many of us are here because we fired up MAME as kids and discovered all those games. Some of them we loosely remembered having played in the long-dead game centers. I would never have been able to come back to them and develop such a strong interest if it wasn't for MAME. A huge number of people (who don't necessarily post on forums) know shmups and arcade games from MAME and play regularly thanks to it.

Especially not with words that have nothing to do with reality.

Unfortunately, they do have to do with reality. I do have to wonder why you defend ShmupMAME so intensely, considering you're not related to the project at all, and it provides such minimal benefit.

Because it is the fruit of voluntary work, produced and distributed for free, allowing us to play the games with additionnal functionnalities that seem indispensable although we never had them before, in nearly perfect conditions that ridicule professional ports, on any PC with any controller. It is proof that the industry only makes the shit that makes them money, whereas passionate people create to make the best thing, for free.

[Op Intensify]

Prometheus, I crown you as the Karl Marx of shmups.

[PROMETHEUS]

Prometheus, I crown you as the Karl Marx of shmups.

:)

[Op Intensify]

But ye of little faith, if it was actually true that the industry only ever "makes the shit that makes them money," most of the games we talk about on a regular basis here would never have been made in the first place. Yagawa would never have been welcomed into Cave, they would have gone completely social/mobile a long time ago, localization companies like Rising Star would never take risks on these games, etc.

Even back in the later Toaplan days, they were going completely against the curve by continuing to develop hardcore shmups while the rest of the industry had fighting game mania. Garegga and Ibara would have been shot down at the first meetings, and Psikyo would have made nothing but strip mahjong after 1996 or so.

It's definitely truer than ever that the industry is greedy and risk-averse, but there are still great original things coming out of it.

[PROMETHEUS]

But ye of little faith, if it was actually true that the industry only ever "makes the shit that makes them money," most of the games we talk about on a regular basis here would never have been made in the first place. Yagawa would never have been welcomed into Cave, they would have gone completely social/mobile a long time ago, localization companies like Rising Star would never take risks on these games, etc.

Even back in the later Toaplan days, they were going completely against the curve by continuing to develop hardcore shmups while the rest of the industry had fighting game mania. Garegga and Ibara would have been shot down at the first meetings, and Psikyo would have made nothing but strip mahjong after 1996 or so.

It's definitely truer than ever that the industry is greedy and risk-averse, but there are still great original things coming out of it.

I don't mean to say that going for maximum profit produces only very bad things, they're just often sub optimal and you will see how money damages the games (like with film, music, and pretty much all other forms of art today).

We have had a pretty shameful example of that this year with Diablo 3. I am disgusted by this piece of vomit and the way the company lied and mislead its customers from beginning to end. Hypocrisy reached levels I had never wtinessed before from a game company, I don't know if it can possibly get worse.

But sure, big companies may produce good games. I like GTA, Doom 3, Half Life 2 and Street Fighter. I like Super Mario 64, and of course I like Dodonpachi !

Now Cave, they make their decisions based on profit like every other company. Look at the way Cave designs their games. The whole business model goes : shmups are games that are not played 100% of the time in the game centers (so it's ok if games last long), but we can try to get the players to play an almost infinite number of runs. So we get : scoring depends on sine qua non perfect execution of an easier first loop (more to memorize too), exponential point system (one less tiny mistake = a lot more points, so you need an ultra perfect game, which multiplies the number of tries before your score is satisfying), TLBs with nearly impossible final patterns, no level select system to properly learn the game (which existed in older games like Pulstar although it was purposely incomplete there)... Look how far they tried to push this model with DOJ white label. No wonder the players said "STOP" and Cave stepped back.

Also, let's be honest. Since Dodonpachi, Cave has experimented with a few new ideas and perfected their one style but overall, they've kinda been remaking the same game 15 times. Less so than Psikyo used to, but it's still striking to me, especially with this last DDP SDOJ which frankly brings absolutely nothing new to the table (although I still like it because the mix suits my tastes).

[trap15]

I'm always very sensitive with things like input lag, and the difference between my PC setup and a cab with PCB in it I can't even feel at all (I can feel the difference between 1 and 2 frames of input lag). I've heard this sprite buffer stuff before, but how much lag are you talking about ? Have you tested it ? Can you make a demonstrable comparison of PCB/MAME/Console Port ? From my own experience, I can bet that if increased input lag there is on PC, it is so small that you can pretty much call it insignificant.

Depending on your setup, PC hardware/emulation will add a latency of at least 1 frame. There is virtually no way to remove that single frame. Sprite buffers (almost) always add 1 frame as well; sometimes games will add more frames of delay, but those are built into the game and are on the PCB as well. Unfortunately, I have no possible way to do a demonstration of the input delays on PCB/MAME/port, but if I did, I'm betting that my analysis is fairly accurate.

Again, console ports have many "inaccuracies", be it with sound, graphics, or even gameplay mechanics and speed. Do we need to compare ports with MAME on those grounds again ? Do you not realize how much more speed inaccurracy affect playing the game than the most minor graphics glitch that every other guy even notices ?
(That said, I think many ports are good enough, the inaccuracies are minor enough. Which is why it is completely laughable in my view to call MAME inaccurate)

For what it's worth, that speed inaccuracy also comes up if you play MAME with VSync on. If you remove that inaccuracy, you get screen tearing. You have to choose the lesser evil. Obviously for a console port, screen tearing is something you just don't want to happen, so they opt for turning on VSync. The only way to avoid screen tearing AND have the same speed, is to be able to set your monitor's vertical refresh to the game's original refresh rate, which happens to typically be quite difficult. MAME tends to be inaccurate often times. Many of the CPU cores don't count cycles properly or wait-states are not emulated (most importantly, the 68000 core that's used by more than half the games that MAME runs; this issue is particularly obvious in almost every CPS1/CPS2 game), and the sound emulation is pretty mediocre (MAME .99 [base of SM3.0b] was much much worse in this regard, and was downright awful for many sound chips).

Many of us are here because we fired up MAME as kids and discovered all those games. Some of them we loosely remembered having played in the long-dead game centers. I would never have been able to come back to them and develop such a strong interest if it wasn't for MAME. A huge number of people (who don't necessarily post on forums) know shmups and arcade games from MAME and play regularly thanks to it.

This point applies to original MAME, not ShmupMAME. So your original point there is still moot.

Because it is the fruit of voluntary work, produced and distributed for free, allowing us to play the games with additionnal functionnalities that seem indispensable although we never had them before, in nearly perfect conditions that ridicule professional ports, on any PC with any controller. It is proof that the industry only makes the shit that makes them money, whereas passionate people create to make the best thing, for free.

Again, a point about MAME, not ShmupMAME.

[PROMETHEUS]

Well, not only do you only bet on 1 frame (and I don't mind betting against you), but 1 frame is not quite a lot. Did you know Border Down has at least 4 or 5 frames of input lag on its original arcade version ? Just try it, it's easy to feel it. Haven't heard anyone complain much about it. Do you feel the lag that you talk about on MAME/ShmupMAME ? Most people can't notice the 1 frame difference between MAME and ShmupMAME, it was not easy for me to feel it. Does it actually bother you ? I can't even feel the difference with PCB. What purpose does criticising MAME for something you can't even feel serve ?

Sure I get some minor screen tearing if I TATE my screen, but I don't really care much for TATE so most of the time I just play Yoko and I don't see screen tearing. I don't know how to set up my monitor to get the game's refresh rate. I suppose there must be a way to do it, otherwise they'd get the same tearing in arcades (sometimes I have actually seen some as well). Still, like you said, screen tearing is the lesser evil compared to speed inaccuracies, it affects gameplay very little. Plus, the speed inaccuracies on console ports can't be due to VSync only (doesn't VSync introduce additional input lag as well?), there are usually also slowdown inaccuracies in many spots due to the game being reprogrammed rather than emulated.

Now ShmupMAME has a less than ideal sound emulation. So what ? Let us build on that and make it better, it can't be hard since other MAME versions emulate sound very well. I don't know what you're talking about with .99 having terrible sound, I know that WolfMAMEplus99 emulated nearly all the games I played perfectly (I have tried them all in arcades as well, and on quiet home setups. By the way, in arcades you usually CANNOT HEAR SHIT). Also, sound inaccuracies have no incidence on gameplay.

It's not like ShmupMAME is a thing graved in stone.

(lol during all that discussion I'm sometimes reminded of the garbage official DDP ports on PSX and Saturn. Even sound and graphics lost quite a bit in quality there which I omited on my post, because they don't really affect gameplay so they don't matter much. http://shmups.system11.org/viewtopic.ph ... 88#p707688 if you're interested.)

PS : if you are interested, you should work on ShmupMAME trap15. It's people like you that we need who understand the games well and who have experience in programming in the context of emulation to make our tools even better. I would donate for that kind of work no problem, like I donated for this forum's servers or for dumping Ketsui.

[renardqueenston]

(I can feel the difference between 1 and 2 frames of input lag)

Most people can't notice the 1 frame difference between MAME and ShmupMAME, it was not easy for me to feel it.

i'm confused, can you feel it or not?

[cools]

What purpose does criticising MAME for something you can't even feel serve ?

I can both feel, see, and if I so desired measure the difference between MAME and the PCB.

[cstle]

Sure I get some minor screen tearing if I TATE my screen, but I don't really care much for TATE so most of the time I just play Yoko and I don't see screen tearing.

The tearing is there in yoko.

Depending on your setup, PC hardware/emulation will add a latency of at least 1 frame. There is virtually no way to remove that single frame. Sprite buffers (almost) always add 1 frame as well; sometimes games will add more frames of delay, but those are built into the game and are on the PCB as well. Unfortunately, I have no possible way to do a demonstration of the input delays on PCB/MAME/port, but if I did, I'm betting that my analysis is fairly accurate.

The Futari port on 360 lets you select the frame delay.

2 frames (labelled PCB), 1 frame (I guess a less lag inducing vsync method, single/double buffer as opposed to triple buffer maybe?), and 0 frame (labelled tearing)

Because it is the fruit of voluntary work, produced and distributed for free, allowing us to play the games with additionnal functionnalities that seem indispensable although we never had them before, in nearly perfect conditions that ridicule professional ports, on any PC with any controller. It is proof that the industry only makes the shit that makes them money, whereas passionate people create to make the best thing, for free.

Why do you continue to call it "the best thing" when that has been proven false? The cognitive dissonance here is staggering. You're sensitive to input lag, but the unavoidable input lag caused as a result of the nature of PCs and emulation doesn't bother you? What? Console ports are worse because of inaccuracies, but MAME is the best even though it also has similar inaccuracies often to a higher degree, on top of other things that console ports don't suffer from? This makes no sense!

[Hagane]

Don't modern consoles also have several frames of delay by default?

[PROMETHEUS]

(I can feel the difference between 1 and 2 frames of input lag)

Most people can't notice the 1 frame difference between MAME and ShmupMAME, it was not easy for me to feel it.

i'm confused, can you feel it or not?

Too bad you're confused, I think it's pretty self explanatory

[PROMETHEUS]

Why do you continue to call it "the best thing" when that has been proven false? The cognitive dissonance here is staggering. You're sensitive to input lag, but the unavoidable input lag caused as a result of the nature of PCs and emulation doesn't bother you? What? Console ports are worse because of inaccuracies, but MAME is the best even though it also has similar inaccuracies often to a higher degree, on top of other things that console ports don't suffer from? This makes no sense!

And when I read you I think the same thing ! We will never understand each other, cstle.

[Op Intensify]

Don't modern consoles also have several frames of delay by default?

I have yet to see any hard proof for it, but multiple sources have told me that the 360 hardware natively has 1-2 fewer frames of input lag than the PS3. It's definitely the case with certain specific games, like SSFIV.

[BPzeBanshee]

I think the correct opinion is that Shmupmame is preferable for certain games with severe input lag problems (Psikyo, Seibu, Raizing from Garegga onwards) and less than desirable for others (Cave).

The ideal logical conclusion, except I've noticed nothing considered 'less than desirable' for side-effects in games under the Cave driver with ShmupMAME. Play ESPRade for a while, then play it in ShmupMAME and try telling me there's no difference then.

[mancity]

Well iv'e found what the problem was, it was SHMUPMAME.

Using standard MAME, the correct game filters etc come across perfectly through Maximus Arcade.

Only problem now is that I think Maximus Arcade is dead because the forums are all closed (last post 2009) and nobody from the site is responding to my e-mails! All I want to do is pay to register for the full version!

[renardqueenston]

Too bad you're confused, I think it's pretty self explanatory

you said "i can feel the difference" then you said "it was not easy for me to feel it"

that's not self-explanitory

[e_tank]

vsync as used by an emulator and vsync as used by a port are usually not comparable when it comes to causing input latency. by itself using vsync doesn't cause input latency, it's how it's used that can cause it.

for example, gradius and gradius 3 (the arcade versions) are two games that sync to vblank on hardware, yet have no input latency. these games both have, when measured from the time input is read to the point where the games reaction frame has completely been scanned out on screen, a ~1 frame response time by default. this is the (practically) absolute minimum response time a game can have.

emulators are more likely to suffer from (up to) an additional frame of input latency (compared to the original hardware) when using vsync due to the fact that the host system running the emulator has to stay synchronized in time to some degree with the emulated system.

the way the vast majority of emulators are setup is for each frame on the host system, the emulated system is stepped one _whole_ frame worth of time. now, by the time the host system has the image that the emulated system would have scanned out for that frame, the host system itself is typically somewhere into its own scannout phase.

at this point the emulator on the host system has a choice, either setup the image it just got to be scanned out immediately, even though part of the previous frame has already been scanned out on the host system's display (this would be tearing), or wait until vblank and set the frame up to be scanned out on the next frame (vsync).

so, technically when you disable vsync you're not really removing 1 full frame of latency, you're removing a fraction of a frame worth of latency because you'll only get part of the screen that amount of time sooner, not the whole frame. however, if your ship is inside the part of the frame that you did get then as far as the player is concerned 1 full frame of latency was removed. ymmv

ports typically don't have this problem, they can do exactly what the original game did, which is to setup the hardware to draw each frame. this will only introduce added latency if the new hardware has some sources of latency that the original hardware did not have, or if these sources cannot be worked around (for example, while i'm not sure i believe you can work-around/essentially-disable double buffering when programming on the xbox360 at a low level)

there's some other stuff i wanted to post on this topic (what futari's may be doing and the "lagless option", double buffering, input polling rate, why i agree with trap15 that shmupmame in its current form is dead, etc), but i think i've gone on long enough for now..


oh, i want to mention something regarding the ps3 vs xbox360 input lag before i end this long ass post. from what i've read ps3 games do tend to have a frame or two more of input latency than comparable xbox 360 games, though i think it's more due to the fact that ps3 is much harder to program for, as i overheard from someone that works on the ps3 that their gl like libs and stuff are just a fucking mess, so it might have something to do with that.

[PROMETHEUS]

Too bad you're confused, I think it's pretty self explanatory :lol:

you said "i can feel the difference" then you said "it was not easy for me to feel it"

that's not self-explanitory

I think that you understand what it means when someone says they can feel the difference and it was not easy to feel it. It means that it was not easy, but I could feel it. Self explanatory, no ?

[cstle]

I'm always very sensitive with things like input lag, and the difference between my PC setup and a cab with PCB in it I can't even feel at all (I can feel the difference between 1 and 2 frames of input lag)

So are you very sensitive or not?

[PROMETHEUS]

I'm always very sensitive with things like input lag, and the difference between my PC setup and a cab with PCB in it I can't even feel at all (I can feel the difference between 1 and 2 frames of input lag)

So are you very sensitive or not?

From my experience I think so. Obviously that is relative to other people. I have my friends try the same thing to see if they feel the lag, often they don't. Gamers and non gamers. I think I got sensitive to this thanks to playing FPS (first person shooters) on PC, because there's a different thing there that tests your sensitivity in a similar way : whether your fps (frames per second) is higher or lower than 60 (depends on your graphics settings/game and PC perf, most games on consoles run below 60 by the way). When it's lower, you immediately feel a significant loss in your playing performance, because the game does not respond as smoothly and fast as your brain can handle. It makes you mistime things and you get innaccurate information from the screen. Over the years I've become able to feel very quickly (like all other competitive FPS players) whether my fps is below 60 or not. Most casual players don't even notice that the game is lagging when they are at 30 fps (competitively unplayable), I've also seen casual players play at 15 fps and not even notice the game was lagging. And yet, fps has such a huge impact that it would be very hard to win a Quake 3 match against an equally skilled opponent if your fps is 50 and his is 60 or higher (higher than 60 seems to make very little difference). I got sensitive to this enough that when the game is running at 55 fps I can already tell it's below 60, without a fps counter displayed (then turning on the fps counter to check).

When ShmupMAME came out, I tested it and at first I was kinda feeling a difference but was unsure if it was psychological (because I THOUGHT ShmupMAME was faster since I was told it was). Coming back to WolfMAMEplus99 after a few seconds of testing I was not entirely sure whether my feeling was psychological or not. So I played a bunch of hours of Hibachi and other stuff and then came back to WolfMAMEplus99 again. Now I could distinctly feel my ship felt heavier in WolfMAMEplus99. I had learned to take advantage of the lesser lag, and I could feel the slight loss of performance with 1 more frame of lag. It is difficult to feel, because it is very small. You need to be very sensitive to feel it. Can you ?

Casual gamers who can't tell 30 fps from 60 fps certainly wouldn't notice such a very small 1 frame difference of input lag, which is what I verified making my friends try ShmupMAME/WolfMAMEplus99. The most competitive of my friends may feel it.

[Pulsewidth]

(higher than 60 seems to make very little difference).

Are you using a 60Hz monitor? The difference between 60fps and 120fps is very obvious on a 120Hz monitor.

[PROMETHEUS]

(higher than 60 seems to make very little difference).

Are you using a 60Hz monitor? The difference between 60fps and 120fps is very obvious on a 120Hz monitor.

Nah I'm using 100Hz monitor, I can see the difference between 60 and 100+ fps but it affects my playing extremely little. Still nice to take though I guess. Maybe to more experienced Quake gamers it's important to maintain FPS over 100, 125 is supposed to be optimal there due to technical reasons I don't accurately understand. Something about having everything in sync, server communication, physics engine and graphic frames... Might be useless in LAN ?

[Ed Oscuro]

Early Quake III Arena versions had some timing bugs that affected either speed or jump height (I forget which). I've seen some people stating that even recent games "work better" with 120 FPS (but this gets different recommendations...would the CS:GO setting be "fps_max 120" or "fps_max 121" here, since it will show 119 in the first instance on the statistics, and 120 in the second) but unfortunately I have seen no real expert commentary on this. In fact knowledge of the old Quake III Arena bugs seems to be fading away from the 'net, since I can't find any article on it at the moment.

On the other hand, something like game tic rate is easy to understand. Basically, you can modify how many times your client is updating the server and being updated by it in turn, so generally a higher rate should be fine, although it must be tweaked for the specific connection to prevent additional updates being requested in the time you are waiting for the server update. Obviously, not getting the number of updates close to the maximum possible after network latency will cause some problems too - too much reliance on client-side prediction. Either case can see "jumpy" movement if the connection is not completely stable at a certain rate (and it almost never is).

None of that really has anything to do with a well-understood, completely contained offline system like a PC or console played offline where all the timings are (supposed to be) known.