Finding the emulation deficiencies - Vol. 1: CAVE games.

[system11]

That's all very true. And the music in RFjet is not correct (yet) indeed.

Furthermore all of the good mame emulations are identical to the PCB.

Nuff said.

Ostriches.

[TWITCHDOCTOR]

All this talk about PCB owners feeling the "sting" is quite stupid. If you have the money, and you want to collect the PCB's, that should be fine. You'll have no issues of any kind, plus, you'll know its running correctly because there's only one way for it to run. Emu's have too many friggin options to set, and therefore, the game might not look or play EXACTLY like the PCB.

I like emulation, and I think its pretty close to the real thing(mybe 98 percent)...but I'm positive that if I owned the PCB, I would be much happier.

BTW: I emulate through an XBOX with composite video. Maybe if I had a "super-computer" things would look and play better...but the cost of a computer, honestly, I would rather buy an arcade cab...I plan to do so, one day. Everything plays fine on the Xbox, but I notice some control "lag" when playing MAME0X stuff. Even though its only slight, it can still feel a little "floaty" at times.

[Recap]

Ostriches.

The truth is that many systems don't have *perfect* emulation today, SPI being one of them. If I recall, besides sound, alpha blending and sprite priorities were not fully emulated on this driver, but that would belong to another thread. Some games, thoe, are exactly the same than the real thing. As I said, I made side-by-side test with the Neo-Geo hardware, a system which I've owned since 1992. And there were not noticeable differentiations at all, including controller/response issues. Maybe we're on the way to say the same for the older Cave games.

Now let us get this topic back on it's rails, the issue was how can we supply the Guru with a PCB that he can use to measure the timing in the blitter chip.

What he said.

I'll be updating the first post with the new info, since I think all the opinions may be useful for the MAME team and also for the ones here who want to know.

The boss battles (and some mid-boss battles) are NOTICEABLY slower on the PCB! At first I thought maybe my board was having problems, but then whenever the amount of enemy bullets lightens up, things speed back up. In fact, I'd almost say that from the begining of the first stage, things seems a tad faster than mame.

Which version/build of MAME are you using?

Anyway, with the timing all different, its like starting over again, except I'm better at the boss battles now

That was another one of this thread's points. Some people thinking about purchasing certain PCB's or with the possibility of playing it at arcade centers most likely won't be playing MAME versions in order to avoid the this-is-not-like-it-used-to-be syndrome now they know. Some others will prefer to wait until this driver gets a revision in order to see if the issues are solved and save the money.

I also got ESPRade PCB, but haven't playing that one as much in Mame to make a good comparison.

Please, do it and let us know.

Look at Giga Wing on Dreamcast- it's got an appreciable speed difference from CPS2.

Nebula's emulation of GW is almost the same than the PCB, on the other hand.

Going back to Dangun, I think there was some speed difference playing in MAME. Then again I was playing the PCB on a 13" commodore and MAME on a 19" flatscreen. I felt is was faster, and in that game a little more speed hurts. The rumor I heard was 54hz pcb vs 58 hz MAME. All I've seen are anecdotes, is there any proof?

Need to know the version/build of MAME you used. The 54 Hz thing is false.

[Venom]

Ah, good to know. I was using MAME32+ 0.83, and a wolfmame around the same version. I thought it was faster as I would practice parts on the emu then move to the PCB and it seemed easier, I don't know

[system11]

Going back to Dangun, I think there was some speed difference playing in MAME. Then again I was playing the PCB on a 13" commodore and MAME on a 19" flatscreen. I felt is was faster, and in that game a little more speed hurts. The rumor I heard was 54hz pcb vs 58 hz MAME. All I've seen are anecdotes, is there any proof?

The core CPU speed and sound CPU speed is 16mhz (probably some associated chips run at this speed too), the rest of the PCB including the output section is 56mhz.

[undamned]

...The rumor I heard was 54hz pcb vs 58 hz MAME. All I've seen are anecdotes, is there any proof?

Need to know the version/build of MAME you used. The 54 Hz thing is false.

Wow, I can't even think of a device that runs that slow!


Just to let you mame kids that are frothing with demand for propper cave emulation know, I PM'd the Guru, with an offer to lend him one of my behbess, yesterday. We'll see if he responds, and if so, I'll be sure and let ya'll know how it turns out.
-ud

[AWJ]

Need to know the version/build of MAME you used. The 54 Hz thing is false.

Wow, I can't even think of a device that runs that slow! :D

That would be the vertical frequency of the monitor (i.e. the framerate), not the CPU speed. Vertical frequencies from 50Hz to 60Hz are typical for arcade games (a standard-resolution arcade monitor is basically a television CRT, and TVs run at either 50Hz or 60Hz).

The "insufficient slowdown" problem is already known by the MAME developers. The reason for it is that MAME's 680x0 CPU core currently does not emulate memory wait states. The upshot is that even though the emulated CPU is running at the correct frequency, it is still "too fast" because instructions execute in fewer cycles than they would on a real board.

The vast majority of 680x0-based games in MAME still run at the correct speed most of the time because the speed of the gameplay in a video game is typically regulated by the monitor's vertical frequency. However, MAME may continue to run at full speed in places where a real board would slow down, because the emulated CPU is more cycle-efficient than the real thing.

Correcting the CPU timing is not tricky to do, it's just an enormous amount of work due to the sheer number of games involved. The speed of the ROM and RAM chips used by each individual game or system would have to be researched in order to determine the correct timings.

[Recap]

Just to let you mame kids that are frothing with demand for propper cave emulation know, I PM'd the Guru, with an offer to lend him one of my behbess, yesterday. We'll see if he responds, and if so, I'll be sure and let ya'll know how it turns out.
-ud

"mame kids"?

Does the fact of collecting used coin-operated artifacts always have the side effect of making you feel like the next best man or is it just a problem with this guy's real life?

[Dave_K.]

The boss battles (and some mid-boss battles) are NOTICEABLY slower on the PCB! At first I thought maybe my board was having problems, but then whenever the amount of enemy bullets lightens up, things speed back up. In fact, I'd almost say that from the begining of the first stage, things seems a tad faster than mame.

Which version/build of MAME are you using?

Mame 0.90 - command line windows pre-compiled binary

I would also like to mention one more emulation oddity, and that is Mame reports ESPrade and DoDonPachi as having stereo sound, yet the boards are mono. It could be that the sound chip is capable of outputing stereo, but is only wired up in mono per the jamma spec? Anyway, I really couldn't tell a difference when listening in mame (other than having better amplified speakers).

[AWJ]

I would also like to mention one more emulation oddity, and that is Mame reports ESPrade and DoDonPachi as having stereo sound, yet the boards are mono. It could be that the sound chip is capable of outputing stereo, but is only wired up in mono per the jamma spec? Anyway, I really couldn't tell a difference when listening in mame (other than having better amplified speakers).

Plenty of games in MAME claim to have stereo sound but don't. Gradius 2 is another example that comes to mind. I think the reason is exactly as you hypothesize.

[undamned]

Just to let you mame kids that are frothing with demand for propper cave emulation know, I PM'd the Guru, with an offer to lend him one of my behbess, yesterday. We'll see if he responds, and if so, I'll be sure and let ya'll know how it turns out.
-ud

"mame kids"?

Does the fact of collecting used coin-operated artifacts always have the side effect of making you feel like the next best man or is it just a problem with this guy's real life?

It's all about first hand experience, Recap. You sound like you would be a great person to talk to about Neo emulation, as you've had much experience with the real hardware to legitimately compare an emu to. On the otherhand, you would not be the person to talk to about Cave emulation, as you lack the appropriate exposure to the real hardware to have any authority on the matter.

My frustration is with those who claim some etherial authority on what is being emulated properly, or not, based purely on what they've read on the internet, half of which is bogus, and not first hand experience. That, Recap, is my problem.
-ud

[Recap]

It's all about first hand experience, Recap. You sound like you would be a great person to talk to about Neo emulation, as you've had much experience with the real hardware to legitimately compare an emu to. On the otherhand, you would not be the person to talk to about Cave emulation, as you lack the appropriate exposure to the real hardware to have any authority on the matter.

My frustration is with those who claim some etherial authority on what is being emulated properly, or not, based purely on what they've read on the internet, half of which is bogus, and not first hand experience. That, Recap, is my problem.
-ud

Nope. Your problem is that you're too prejudiced to even read properly my posts here. Of course it's all about "first hand experience". That's why I was asking to those who have it with the Cave hardware. I never claimed any "etherial authority" on Cave emulation besides that it is currently not perfect. You're totally missing the point. Still.

[system11]

Nope. Your problem is that you're too prejudiced to even read properly my posts here. Of course it's all about "first hand experience". That's why I was asking to those who have it with the Cave hardware. I never claimed any "etherial authority" on Cave emulation besides that it is currently not perfect. You're totally missing the point. Still.

Your problem is you've approached the whole subject with a faulty assumption, steadfastly refused to accept the truth from people who actually know, and now you're calling the one guy who offered to send a PCB to Guru, prejudiced.

You sir, are skipping merrily along the lunatic fringe.

[undamned]

I'm pretty sure I'm casting pearls before swine, at this point, but I said I'd share, so here's his reply to my questions/offer:

Hello,
Correcting any issues involves running the PCB side-by-side with the emulation and noting the differences. Then using some equipment to measure a few frequencies that the PCB outputs. It can also involve running some custom code on the PCB to measure things internally.
...
Time frame is from 1 month to 10 years. I have no control over that because it depends on the other devs, who may or may not have time or motivation to look at it.
Your PCB is pretty standard as far as PCBs go, so there's no possibility of damage from what I would be doing.

Guru

-ud

[Recap]

Your problem is you've approached the whole subject with a faulty assumption, steadfastly refused to accept the truth from people who actually know, and now you're calling the one guy who offered to send a PCB to Guru, prejudiced.

"Faulty assumption"? That perfect emulation is possible? So _you_ are one of those who "actually know the truth" and doesn't really matter if someone else has made side-by-side comparisons?

Hilarious ego-statements, yours.

[system11]

Your problem is you've approached the whole subject with a faulty assumption, steadfastly refused to accept the truth from people who actually know, and now you're calling the one guy who offered to send a PCB to Guru, prejudiced.

"Faulty assumption"? That perfect emulation is possible? So _you_ are one of those who "actually know the truth" and doesn't really matter if someone else has made side-by-side comparisons?

Hilarious ego-statements, yours.

A lot more side by side comparisons than a couple of Neo carts, yes. This is speaking as someone who used to repair the things for a charge, and had 250 or so in his personal collection (the vast majority of which has now been sold - you don't want to know how much space they take up). MAME can be very useful when you're not sure if something is right or wrong.

You say you want peoples input, we point out issues, we mention that emulation can't be perfect due to certain factors, but you just won't let it go - all because you tested KOF 2001 and it looked about right. That does not = perfect. Perfect involves emulating precisely the speed of the RAM, it means running emulation of every specific chip on the PCB at the exact clock rate it uses on the real thing (some of which a PC clock can't actually do without approximating them). You need to understand the difference between emulating enough of the hardware to run the code, and actually emulating a game board. The former is possible with varied results from hilariously bad all the way up to virtually identical, the latter is not going to happen.

I guess at this point having stated the obvious again (without even going near the fact that almost no PC is equal to another), I'll have been judged to be making "ego-statements". Seems to be the only assessment you have the familiarity to make.

[visuatrox]

I can say it this way: ANY arcade to console port has more behaviour differences than a big part of today's emulated games. Even the ones from Naomi to DC and other analogies.

True, but for me personally it is about the whole experience of playing the game. Even though a console port would not be completely true to the original, for me it feels more "real" than playing the game on an emulator. Atleast the console ports are approved releases.

And then I feel playing for score on an emulator is pointless (speed differences, different versions of software, hardware etc..). On a console port everyone is playing on the same kind of hardware, and skill can then actually be measured correctly.

[Recap]

A lot more side by side comparisons than a couple of Neo carts, yes. This is speaking as someone who used to repair the things for a charge, and had 250 or so in his personal collection (the vast majority of which has now been sold - you don't want to know how much space they take up). MAME can be very useful when you're not sure if something is right or wrong.

You say you want peoples input, we point out issues, we mention that emulation can't be perfect due to certain factors, but you just won't let it go - all because you tested KOF 2001 and it looked about right. That does not = perfect. Perfect involves emulating precisely the speed of the RAM, it means running emulation of every specific chip on the PCB at the exact clock rate it uses on the real thing (some of which a PC clock can't actually do without approximating them). You need to understand the difference between emulating enough of the hardware to run the code, and actually emulating a game board. The former is possible with varied results from hilariously bad all the way up to virtually identical, the latter is not going to happen.

Read again the thread. I've explained for three times that with *perfection* I mean *virtually identical*, that is, WITH NO DIFFERENCES FOR THE MOST EXPERT PLAYER. Or as identical as different PCB's of a certain game are each other, if you prefer. That's what is important for us, the users who are not repairers, and that's this thread's point. "LALALALAL I CANT HEAR YOU vented in text" you said? Funny...


Oh, and believe me, after more than a decade owning the system, I had the chance to test a bit more than "a couple of Neo carts".

True, but for me personally it is about the whole experience of playing the game. Even though a console port would not be completely true to the original, for me it feels more "real" than playing the game on an emulator. Atleast the console ports are approved releases.

See how are all the "approved released" these days - full of "real" filters and upscaled graphics.

And then I feel playing for score on an emulator is pointless (speed differences, different versions of software, hardware etc..).

That's why I think that the *perfect* emulation status is so important. If the software is coded well enough and you have the correct equipment, we wouldn't be speaking about "differences". Anyways, "playing for score" on the real hardware is equally pointless - some are using pads, some are using console sticks, some are using Sanwa sticks, some others, auto-fire functions...

[Dave_K.]

Something else I noticed with my DDP pcb. On the first boss, when he starts doing the slow blue elongated spread fire, there are a lot of bullets that start blinking....so much so that its difficult to keep track of their trajectories!

As for the ESPRade pcb, with Yuusuke's first boss, the entire status bar on the bottom will sometimes completely disapear as the boss is exploding.


I wonder if anyone has compared DDPDOJ and ESPGaluda PS2 ports with the PCB. There aren't any "wait" options that I can find, so I wonder if these are considered same timing wise as the PCB...

[Recap]

Thanks. The Esprade issues were the ones I read before. Besides the non-disappearing sprites, is there less slowdown in the MAME version?

I wonder if anyone has compared DDPDOJ and ESPGaluda PS2 ports with the PCB. There aren't any "wait" options that I can find, so I wonder if these are considered same timing wise as the PCB...

Good question.

[raiden]

Something else I noticed with my DDP pcb. On the first boss, when he starts doing the slow blue elongated spread fire, there are a lot of bullets that start blinking....so much so that its difficult to keep track of their trajectories!

it doesn´t do that on my setup. I suspect it´s rather the power supply adjustment than a genuine property of the PCB, but it could be different production versions as well. On my game, only the stars happen to disappear sometimes, but the bullets for some reason never do.

I wonder if anyone has compared DDPDOJ and ESPGaluda PS2 ports with the PCB. There aren't any "wait" options that I can find, so I wonder if these are considered same timing wise as the PCB...

I had the opportunity to play ESPGaluda in both versions, and I can say that the PCB looks sharper, but slows down a lot more than the PS2 version - similar to the way Giga Wing CPS2 is different from the DC version.

[Dave_K.]

Thanks. The Esprade issues were the ones I read before. Besides the non-disappearing sprites, is there less slowdown in the MAME version?

I haven't played the mame version enought to see yet.

Something else I noticed with my DDP pcb. On the first boss, when he starts doing the slow blue elongated spread fire, there are a lot of bullets that start blinking....so much so that its difficult to keep track of their trajectories!

it doesn´t do that on my setup. I suspect it´s rather the power supply adjustment than a genuine property of the PCB, but it could be different production versions as well. On my game, only the stars happen to disappear sometimes, but the bullets for some reason never do.

I'll check the +5 on the PSU, but I'm sure its fine (the cab is actually very clean and in great condition). I forget what the exact voltage people run with, should it be +5.17V? The PCB is also clean and looks like its brand new...even though I know thats not possible. I wonder if this is just a function of the TTL switching getting older and degrading over time?

I wonder if anyone has compared DDPDOJ and ESPGaluda PS2 ports with the PCB. There aren't any "wait" options that I can find, so I wonder if these are considered same timing wise as the PCB...

I had the opportunity to play ESPGaluda in both versions, and I can say that the PCB looks sharper, but slows down a lot more than the PS2 version - similar to the way Giga Wing CPS2 is different from the DC version.

Darn, this is what I was afraid of (the slow down). I haven't invested in these two PCBs because they have cheap PS2 ports with cool arrange mode extras. I play them using a PS2->Jamma board, so the display is just as sharp as my other PCBs. But if it can't mimick the slowdown, then its just that much harder for an intermediate shmupper like myself.

[raiden]

I'll check the +5 on the PSU, but I'm sure its fine (the cab is actually very clean and in great condition). I forget what the exact voltage people run with, should it be +5.17V? The PCB is also clean and looks like its brand new...even though I know thats not possible. I wonder if this is just a function of the TTL switching getting older and degrading over time?

I don´t have a clue, but I will go back and check again, maybe I just didn´t notice it yet. My power supply is adjusted to 5.0V, as exactly as I was able to with an analog voltmeter, but this was measured without a PCB attached, and depending on the PSU, the voltage can vary depending on power being drawn.

Darn, this is what I was afraid of (the slow down). I haven't invested in these two PCBs because they have cheap PS2 ports with cool arrange mode extras. I play them using a PS2->Jamma board, so the display is just as sharp as my other PCBs. But if it can't mimick the slowdown, then its just that much harder for an intermediate shmupper like myself

what I meant with sharpness is nothing like the difference between picture signals, but the sharpness which gets lost by scaling to a different resolution. Afaik, the PS2 doesn´t have the exact same resolution as the PCBs do. The scaling is being smoothed by a filter, of course, but it´s not quite as sharp as a picture in its original resolution where pixels can be made out much more clearly.
I´m still in the process of deciding between a PS2 for ESP Galuda or the PCB, and I can´t really come to a conclusion. I like the arrange mode very much, and of course there are many other cool games on PS2 (most of which I don´t really need), but I wasn´t planning to use a hardware with moving parts again if it can be helped, and like you, I suspect the game to be easier on PCB, as is the case with DDP and Gigawing.

[Neon]

raiden: OT I know, but is the PSX DDP harder even with 'wait on' enabled? I was thinking I'd go with that port instead of the PCB simply because it's cheaper yet supposedly arcade perfect...

[Recap]

Something else I noticed with my DDP pcb. On the first boss, when he starts doing the slow blue elongated spread fire, there are a lot of bullets that start blinking....so much so that its difficult to keep track of their trajectories!

it doesn´t do that on my setup. I suspect it´s rather the power supply adjustment than a genuine property of the PCB, but it could be different production versions as well. On my game, only the stars happen to disappear sometimes, but the bullets for some reason never do.

I'll check the +5 on the PSU, but I'm sure its fine (the cab is actually very clean and in great condition). I forget what the exact voltage people run with, should it be +5.17V? The PCB is also clean and looks like its brand new...even though I know thats not possible. I wonder if this is just a function of the TTL switching getting older and degrading over time?

If this thread serves to elucidate that indeed different PCB's of the same Cave game have noticeable different behavior, I can see some emus-are-for-kids theories here falling apart soon. Anyways, are you sure you are not speaking about different DDP revisions?



By the way, how do you connect your PS2 to an arcade (non-scart)monitor?

what I meant with sharpness is nothing like the difference between picture signals, but the sharpness which gets lost by scaling to a different resolution. Afaik, the PS2 doesn´t have the exact same resolution as the PCBs do. The scaling is being smoothed by a filter, of course, but it´s not quite as sharp as a picture in its original resolution where pixels can be made out much more clearly.

Do you play your PS2 in a 'tated' RGB TV? I haven't noticed any scaling or filtered picture in Arika's ports in tate mode, and that was indeed these ports' best acclaimed feature. Haven't played them recently, thoe.

[Dave_K.]

what I meant with sharpness is nothing like the difference between picture signals, but the sharpness which gets lost by scaling to a different resolution. Afaik, the PS2 doesn´t have the exact same resolution as the PCBs do. The scaling is being smoothed by a filter, of course, but it´s not quite as sharp as a picture in its original resolution where pixels can be made out much more clearly.

The PS2 Cave ports are not scaled, this was discussed in a different thread. So the only real issue is the lack of wait/slowdown.

If this thread serves to elucidate that indeed different PCB's of the same Cave game have noticeable different behavior, I can see some emus-are-for-kids theories here falling apart soon. Anyways, are you sure you are not speaking about different DDP revisions?

I don't know what indicates the different revisions of the PCB. I'll have to examine my PCB more closely for any markings and compare with Raiden.

By the way, how do you connect your PS2 to an arcade (non-scart)monitor?

Easy, hack a scart cable with LM1881 circuit to output RGBHV and Gnd. I also have a Playstation2->Jamma converter board (but may be selling it off soon).

[Recap]

Easy, hack a scart cable with LM1881 circuit to output RGBHV and Gnd. I also have a Playstation2->Jamma converter board (but may be selling it off soon).

LM1881 circuit? Please, elaborate.

[matt]

The LM1881 is a popular sync stripper IC. Asking in the hardware section (or Google) should give you plenty of info.

I'm surprised you don't know about it already, given the nature of most of your posts.

[Dave_K.]

Easy, hack a scart cable with LM1881 circuit to output RGBHV and Gnd. I also have a Playstation2->Jamma converter board (but may be selling it off soon).

LM1881 circuit? Please, elaborate.

More info here on LM1881 sync splitter:
http://www.gamesx.com/misctech/lm1881.htm

And here is the console FAQ I wrote for arcadecontrols:
http://arcadecontrols.com/arcade_consoles.shtml

[Dave_K.]

My DDP PCB says "1997 2/5 MASTER VER." at the bottom of the bootup Japan warning screen. With the board powered, I'm reading +5.23V.

I mispoke on where the slowdown is. Its actually the second boss, when he starts throwing out the slow blue elongated fire. But it only starts blinking and everything slowing down while I hold the A button down. If I release it, everything speeds back up and no blinking. I use the first (red) ship with laser type secondary fire.