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

[Kaiser]

Instead of whining over everything random and forgetting important facts. Let's focus on FIGURING out what is causing those DEP errors on modern OSes and how to fix them permamently! Haze, I don't like what you're posting either, the fact is, you're not helping nimitz but rather, trying to bring him down because he's not doing stuff according to your made-up programming MAME ideologies. I'll say it again like someone did before, This is SHMUPMAME, not EVERYTHINGMAME. Shmupmame is nimitz's build and he can do whatever he wants with it. You can't do ANYTHING about it sir. This is his build, not yours! If you don't like it Haze, do us a favor and leave instead of causing more trouble like a rude troll in a popular disguise.

C'mon people, forget Haze. Let's move on and help nimitz with those DEP errors as he needs our support on that.

[PROMETHEUS]

For me this 3.0b version launches properly on this computer (unlike 2.2 which couldn't launch most of the time), but savestates do not work properly (they cause the game to switch to a different version of the good old back label universe, most of the time). Tried in DDP DOJ white and black label. I also still get one very minor graphics innacuracy (contour flickering on a few parts of backgrounds). Sound is now working perfectly though.

[gatsu25]

I've been getting into using Mame finally and using this version of it as well as mameuifx32. I keep having the same problem in both though, and I can't find anything on what I can do to fix it. My controls aren't saving. When I press tab, and set my controls to my Hori stick, it works great, but often it doesn't work the next time I play the game, and I have to reset them. The strange thing is, sometimes it does save them, and I don't need to reset it. Does anyone have any idea what I can do so that it always saves my configuration?

[nimitz]

Is the background flicker only when panning? if so that's due to making the driver lagless.

I did implement a rudimentary save-state support, but i think i'm missing a variable, this will be fixed in the next version.

As for recording, you currently need to set the game to "defaults" (in the test menu) at the beginning of your recording.

[PROMETHEUS]

Is the background flicker only when panning? if so that's due to making the driver lagless.

I think so, will try it out later tonight and tell you if it's not. If it's due to the lagless driver, this issue is very minor anyway, there are only a very few parts of the game where it is visible, and the flicker doesn't bother, just makes the game look a tiny bit worse than it should.

I did implement a rudimentary save-state support, but i think i'm missing a variable, this will be fixed in the next version.

Oh ok that's cool I'll be waiting for the next version then.

[autumndrone]

Did you enable "joystick input" in the options? IIRC it was not enabled by default.

Yes joystick input is/was enabled, no dice.

[austere]

Let's focus on FIGURING out what is causing those DEP errors on modern OSes and how to fix them permamently!

To find out what's happening, you'd need to run it in a debugger (e.g. MSVS Debugger, ollydbg etc.) and study the stack after it crashes or set breakpoints at some suspicious locations around buffers. Even then, it might be triggered by an earlier problem, making it very difficult to track down. Unless you can do this, discussing it any further will prove fruitless. nimitz will not benefit from any more complaints on this particular issue -- in fact, it'll just waste his time.

If you don't like it Haze, do us a favor and leave instead of causing more trouble like a rude troll in a popular disguise.

Don't get disrespectful and turn this into showdown. You guys are very lucky to have someone like Haze give you such detailed information (some of which might lead to some optimisation in the PGM video driver). Anyway, Haze's trolling was pretty classy.

[IseeThings]

Let's focus on FIGURING out what is causing those DEP errors on modern OSes and how to fix them permamently!

To find out what's happening, you'd need to run it in a debugger (e.g. MSVS Debugger, ollydbg etc.) and study the stack after it crashes or set breakpoints at some suspicious locations around buffers. Even then, it might be triggered by an earlier problem, making it very difficult to track down. Unless you can do this, discussing it any further will prove fruitless. nimitz will not benefit from any more complaints on this particular issue -- in fact, it'll just waste his time.

If you don't like it Haze, do us a favor and leave instead of causing more trouble like a rude troll in a popular disguise.

Don't get disrespectful and turn this into showdown. You guys are very lucky to have someone like Haze give you such detailed information (some of which might lead to some optimisation in the PGM video driver). Anyway, Haze's trolling was pretty classy.

I wouldn't call it trolling anyway, it's a build called ShmupMAME where quite a few Shmup drivers are in a worse state than they are in current official builds while claims are being made by the developer about how much better the build is. I think it's only fair to point out a few examples, what the devs of this build do with them is up to them Also I see no harm in saying it's not going to be the best build for creating YouTube videos, the lagless changes do cause video artifacts, that's undeniable; if it wasn't the case we wouldn't even know the hw was meant to sprite buffer in the first place.

DEP errors? IIRC it might have been something to do with the video blitters, but I honestly can't remember that far back. There was definitely a related fix made at some point, but it might have been part of the video rewrite you're trying to avoid....

[Cugel]

Nimitz: Thanks for your extraordinary work!

To reiterate what others have said: This is a very focused MAME build. It's for reduced lag shmups. If you need to run Pac-Man, a general MAME build is what you need. I run ThunderMame32+ for Toaplan shmups with sound samples. GareMame for Battle Garegga.

I'm still having problems with the v3.0b build. DEP exception doesn't work. And I have no problems running emulators such as PCSX2 or SSF on my Win 7 64-bit system. Any chance you can make a 64-bit build?

[StarCreator]

Quite honestly, the thing I find most appalling about this thread is people actually shutting down DEP to make a pirated video game work. That's kind of like removing the doorknob from your front door because a copy of your housekey you got at the hardware store didn't fit into the slot.

I can count how many things I'd shut DEP down for on a hand with all of its fingers cut off.

[Barrakketh]

Quite honestly, the thing I find most appalling about this thread is people actually shutting down DEP to make a pirated video game work. That's kind of like removing the doorknob from your front door because a copy of your housekey you got at the hardware store didn't fit into the slot.

No, it's more like giving a copy of your house key to a friend so he can feed and walk your dog while you're on vacation.

You see, it's a DEP exception. Do you know what the word "exception" means? In this case it means you're not using DEP for MAME...you know, like that friend you let have a key to your house so they can walk your dog. Not shutting it off completely.

[TrevHead (TVR)]

Quite honestly, the thing I find most appalling about this thread is people actually shutting down DEP to make a pirated video game work. That's kind of like removing the doorknob from your front door because a copy of your housekey you got at the hardware store didn't fit into the slot.

There is a big difference between making a DEP exeption for one program and turning DEP off completly which is a very stupid thing to do.

EDIT I see someone bet me

[StarCreator]

No, it's more like giving a copy of your house key to a friend so he can feed and walk your dog while you're on vacation.

You see, it's a DEP exception. Do you know what the word "exception" means? In this case it means you're not using DEP for MAME...you know, like that friend you let have a key to your house so they can walk your dog. Not shutting it off completely.

You're completely missing my point. If we go with your analogy, it's more like giving your housekey to some random guy on the street. No offense to nimitz, but I don't trust anything enough to make it a DEP exception - ESPECIALLY if he can't tell me WHY it needs a DEP exception.

Sorry for the bad semantics, though. Pulling work-at-home all nighters can do that to you.

[TrevHead (TVR)]

The way I see it is that if v3 really was dodgy it wouldnt be an intermitant fault on some PCs but flag everytime on everyones PC that was XP sp2 and above.

I C+P this from one of those help websites
Some software/application behaviors are incompatible with the data execution prevention. Applications which perform dynamic code generation (such as Just-In-Time code generation) and that do not explicitly mark generated code with Execute permission might have compatibility issues with data execution prevention. Applications which are not built with SafeSEH must have their exception handlers located in executable memory regions

means fuck all to me I just wanna look like I know what im talking about

[Teufel_in_Blau]

I tried Shmupmame v3.0b and the sound works finally perfect during the bossfights and in the middle of Stage 4 in Ketsui.

But, when I'm at the start of a new stage, after you got your points, my helicopter disappears for one or two seconds. Also during the last pattern of the last boss, it makes a weird sound, something like HUSH HUSH...

Maybe it's just my rom... Anyone experienced the same thing?

[Barrakketh]

Applications which perform dynamic code generation (such as Just-In-Time code generation) and that do not explicitly mark generated code with Execute permission might have compatibility issues with data execution prevention.

[...]

means fuck all to me I just wanna look like I know what im talking about

Actually, that is probably relevant. From this page on mamedev.org:

In order to address this, a second, more aggressive approach is commonly used: the dynamic recompiler (often abbreviated as dynarec or drc, and synonymous with just-in-time or JIT compilation).

So that might be an issue, but wouldn't explain why people who've made a DEP exception are getting the crash.

I tried Shmupmame v3.0b and the sound works finally perfect during the bossfights and in the middle of Stage 4 in Ketsui.

But, when I'm at the start of a new stage, after you got your points, my helicopter disappears for one or two seconds.

From a few pages ago:

3.0b : ketsui : after stage score counting :
player and enemies disappear from screen for like 2 seconds.
quite a problem on earlier stage 3 where theres a lot of enemies onscreen.

Not hard to fix, nimitz just has to update the end of sprite marker code.

[Teufel_in_Blau]

Ah, thank you. I didn't looked that far behind, my fault.

[TrevHead (TVR)]

So that might be an issue, but wouldn't explain why people who've made a DEP exception are getting the crash.

Speaking of crashes v3 decided to start up again but wouldnt run any roms when I tried starting Mars matrix with v1.5 it BSOD (I should of copied down what was on the screen, but i didnt, it did say it was caused by the memory though) After rebooting I tried playing MM on vanilla wolfmameplus99 and that wouldnt work aswell even after several reboots (1st time this has ever happened).

So ive removed the DEP exception and ran CC cleaner for the registry, planning to give up on shmupmame99 and just stick with wolfmame99 and SM v2.2. Funny thing is, is that v3 / v1.5 have started working again. Fucking sods law isnt it

Hopefully someone who has the fault can run a debugger like Austere suggested. Im not cos im not tech savvy enough when it comes to programming, debuggers and all that jazz.

[evil_ash_xero]

A few things I have noticed about 3.0b. One, in ESPGaluda, sometimes the green gem number will disappear from the screen.

The other thing, is sometimes when I shut MAME down, the sound will continue. Like, the game is still running, but there is no picture. I have to go into my Task Manager and shut down about 4 MAME processes to get it to stop.

Also, DOJ Black Label doesn't show up in my "Available" folder anymore.


Keep up the good work though, this is awesome overall.

[BPzeBanshee]

A few things I have noticed about 3.0b. One, in ESPGaluda, sometimes the green gem number will disappear from the screen.

The other thing, is sometimes when I shut MAME down, the sound will continue. Like, the game is still running, but there is no picture. I have to go into my Task Manager and shut down about 4 MAME processes to get it to stop.

Also, DOJ Black Label doesn't show up in my "Available" folder anymore.


Keep up the good work though, this is awesome overall.

For your DOJ issue, audit DOJ Black Label by itself and see what it tells you. The namings for it have been changed around a few times, if that's the case PM me and I'll help you out.

As I recall, the green gem number has already been reported, so dont worry about that it's not just you.

As for the shutdown thing, how are you shutting MAME down? Do the games themselves run well when playing?
I just use Escape key on the actual game, usually it instantaneously takes me back to the game selection folders but on lesser machines with more taxing games it might take a little bit. Doing multiple things at once while running a game via MAME could possibly cause issues on your machine too.

[PsikyoFan]

Words cannot express my reaction to some of the misconceptions in this thread. What started as an interesting discussion way back in http://shmups.system11.org/viewtopic.ph ... 4&start=47 has snowballed into one of the lowest signal to noise I've ever seen and I feel the urge to apologise unreservedly for my creation (although if it leads to austere and perhaps nimitz/trap15 making more substantial contributions to MAME in the future then that's cool).

I feel the urge to reiterate this again. ShmupMAME intentionally degrades the emulation, it does so to reduce the delay in the rendering of sprites, and so the response times from input to display. THIS DELAY WAS ON THE ORIGINAL MOTHERFREAKING HARDWARE, and while this is an easy way to reduce the lag (and will do so consistently across PC setups), you should probably be looking at other sources of lag first:
* Your display device (LCDs > CRT, especially with fancy post-processing or rescaling)
* Input drivers (USB typically > PS/2, DB9)
* Video drivers

It can also only do so much (a single frame in many cases, two at most). Is it really worth the desync from the backgrounds (I beleieve that it's a net positive for Tetris with its unrelated backgrounds, but just a curio for shmups)? This thread is as much about 'sticking it to the man' and 'making MAME cool again' as it perceptible improvements. No-one has presented any evidence of consistently higher scores resulting from using this that I've seen and proppsed on page 1 of this thread (do correct me if I'm wrong though, I certainly do believe that a reduction in lag of a single frame should lead to an improvement, I'm just not certain it should be perceptible).

As to the crap about MAME becoming bloated and slower, or C++ making a mess of the codebase. These things have been disproved time and again. The code's never been in better shape, and for anything using >256 colours, it's never been significantly faster (more than a few percent). Yes, for games with a small palette they did slow down substantially back around 0.100, but if pacman dropped from running 10000% to 8000% on your desktop who gives a crap? Plus, the improvements gained from shifting to 64-bit and modern CPU cores/compilers shouldn't be underestimated.
And as for the DEP errors, it is likely just a buffer overrun somewhere, or as Haze noted, something to do with the video blitters. It's not got anything to do with the dynarec CPU cores although the buffers used for them do have to be flagged fo execution (And are, Aaron Giles works for MS so this was fixed at the first opportunity...).

[Estebang]

I'm not up on the technical/programming background of MAME, but I am absolutely sure that the ships in lagless shmups seem more agile and fluid. In games that had especially bad lag, it's incredibly noticeable how much easier they are, specifically Psikyo shmups.

Though it would be nice if someone could let me know why my version of Ketsui that worked fine in the last version is rejected by 3.0.

[SPYvsSPY]

Though it would be nice if someone could let me know why my Ketsui rom that worked fine in the last version is rejected by 3.0.

Use 3.0b. it's working perfect for me, though I have to turn on the sync to monitor or the sound is off. Try getting the rom from a different site?


Sorry! sorry.. it's my fault that the C++ thing got outta hand? I should have never brought it up in the first place as I hadn't tried a 139 mame and up as of yet. I've since tried 141 and I must say that my CPU usage in the same games is way down(48-54% in 3.0b vs 26%-36ish in 141 mame) and it runs smooth as silk on my PC setup.

I think it would be impressive to have a modern mame laggless build. Anyways, it's nice of "nimitz" to share his custom build of mame with the world and I'll take whatever version he releases.

[PROMETHEUS]

It can also only do so much (a single frame in many cases, two at most). Is it really worth the desync from the backgrounds (I beleieve that it's a net positive for Tetris with its unrelated backgrounds, but just a curio for shmups)? This thread is as much about 'sticking it to the man' and 'making MAME cool again' as it perceptible improvements. No-one has presented any evidence of consistently higher scores resulting from using this that I've seen and proppsed on page 1 of this thread (do correct me if I'm wrong though, I certainly do believe that a reduction in lag of a single frame should lead to an improvement, I'm just not certain it should be perceptible).

I could indeed feel improvement in ease to dodge patterns and be more accurate in general in DDP, on which I can easily notice improvements because I know exactly how well I perform and how often, everywhere. I also did improve my score very significantly after the lagless version was released (score increase and large rate of success improvement from the stats I've done on my runs, which can be summed up as : reached it in 38 tries instead of 275 last time). It is still difficult to tell if this improvement is due to the reduced input lag or because I've gotten better, I think it's probably both, but I haven't played a lot of DDP between the two series of tries so I think the reduced lag is definitely significant in that improvement. Luck also plays a role, technically it's always possible that you succeed on any try, so if you're very lucky you might get your target score in the first run and think you've had huge success rate improvements, for example.

[Estebang]

Though it would be nice if someone could let me know why my Ketsui rom that worked fine in the last version is rejected by 3.0.

Use 3.0b. it's working perfect for me, though I have to turn on the sync to monitor or the sound is off. Try getting the rom from a different site?

I am using 3.0b. It says it's missing "b04701w064.u1".

If there haven't been any new versions of the rom, there's nothing I can do.

[Treasurance]

If the delay was on the original hardware...that means using ShmupMAME 3.0 for those games is cheating and scores made with it are invalid

[PsikyoFan]

If the delay was on the original hardware...that means using ShmupMAME 3.0 for those games is cheating and scores made with it are invalid

I've made that point before. See http://shmups.system11.org/viewtopic.ph ... 4&start=47.

[TVG]

If the delay was on the original hardware...that means using ShmupMAME 3.0 for those games is cheating and scores made with it are invalid

It seems believable to me. Considering you are using a CRT of course and suffer no lag from other devices.
I know for a fact fighting games never respond on the first frame. I find it odd that shooters do.
2 frame response seems what I would expect on normal hardware.
What puzzles me more are the games that really lag, ie: battle garegga, mars matrix, galaga 88 etc. Anyone knows how they behave on an actual machine?

[IseeThings]

If the delay was on the original hardware...that means using ShmupMAME 3.0 for those games is cheating and scores made with it are invalid

Correct, which is why I told the MARP guys not to accept any replays / scores from this build. If people think it's giving them an advantage, it's an unfair one.

Also if you lag-free certain games like Gratia this way you can see exactly which parts of the scenery can be destroyed to reveal powerups because they're sprites which lag behind the scrolling.

If people like it, fine, but be under no illusions that it is _broken_ emulation, designed specifically to give you an advantage.

Some games have lag built in, it's in the game code, you can study the code, disassemble it, you have all the proof you need. Anything running on Megadrive hardware for example (Thunder Force AC) is capable of updating sprite positions on the very next *line* when rendering, there are no sprite frame buffers, any lag in those games is programmed into the game code. In other cases it's part of the hardware.

If you honestly feel that MAME / your PC is adding more lag then you should look to be eliminating that properly, rather than breaking the emulation to 'simulate' it's removal.. It's not the same thing, not at all. A real solution would be universal, not hacked on a per-game basis.

[moozooh]

Comparing a bare PCB to a bare game driver in regards to the lag is somewhat misleading. You should take the whole system into account to determine the net difference.

I, for one, don't doubt that the whole chain of <keyboard -> USB controller -> input driver -> emulator -> graphics driver -> graphics card -> monitor> gets an unfair advantage over <arcade stick -> JAMMA input -> PCB -> JAMMA output -> monitor> even with 1 frame removed.