forums.marpirc.net

eah Rick, i doubt you'd want to see all the possible scores "achievable" with mame, you know there are just plain -cheat methods out there that allow you to be invincible etc., and with that you can get as high scores as you want. I think you really only want to see high scores that are legit, which alphamame is trying to do.

Legit yes...but you almost are implying that ALL regular mame scores aren't legit. I'm guessing 99% of those are also legit. To group those 99% all together with the 1% or less of those using cheats or slow speed etc. just isn't fair. for leaderboard points fine...it's fair for that...but if I want to see an inp of the highest score ever achieved in a game I want to be able to easily find that score....whether it's legit or not or with alphamame or not.

It's a shame as it appears German's Frogger score will be overlooked by most here when if a legit score it's really quite an accomplishment!

Yes but the point made above is this Frogger score isn't a new score. It's just trying to reinstate an old MARP score that was removed by someone in the past. A reinstatement of an old score shouldn't be subjected to alphamame blocking cuz it was submitted prior to 3/3/03.

So if you are reaccepting this 600k Frogger score even with the admission of pausing by the player then it shouldn't be subjected to alphamame blocking.

As much as a big issue is made of alphamame blocking here at MARP it must have really been a serious issue at some time for some games. That's a shame.

I guess there are many out there that really aren't true gamers. A true gamer would never cheat. True gamers have integrity and respect for the game.

While I agree for some games the speed it emulates at is very important, for many games it isn't.

Frogger isn't exactly a fast paced game. Playing it at even 50% speed really wouldn't make the game hardly any easier IMHO.

It's all in judging the timing of the moving cars to get across the road and then judging alignment of logs/alligator etc. to get your frog "home".

I don't think that judgement gets any easier with the game running slower....in fact at some point I think it would be harder to time all of that right.

For many other games MARP tracks like mahjong and other quiz/puzzle type games that have some timer involved in awarding points based on speed of response of an answer etc. the game playing at 90+% speed is extremely important...to the point I personally would require it average out to 99+%...instead of MARP's 90%. That extra 10% time can mean a big difference for some of those quiz games. To get a higher score from having it run even 5-10% slower when you could easily add a bit more frameskip to have it run at 100% speed is cheating as well.

It just seems MARP policies there are a tad inconsistent.
If someone gets 92% average speed but using frameskip 0 IMHO that score should be DQ'd cuz they could have easily used a little frameskip to get 100% speed.

Also, I find it odd just cuz BBH for round 1 of K3 can only get 80% the person managing it just arbitrarily then declares...ok, 80% is good enough speed for this. No exceptions, especially personally based ones, should be made for policies like speed.

As far as macmame is concerned, I know for TG I am not going to accept any games where speed of the game really makes a difference in how the scoring can be until there is a version of macmame that has info saved to the inp related to speed. This would include all the mahjong or other card games as well as trivia games etc. at least.

LN2 wrote:Legit yes...but you almost are implying that ALL regular mame scores aren't legit. I'm guessing 99% of those are also legit. To group those 99% all together with the 1% or less of those using cheats or slow speed etc. just isn't fair. for leaderboard points fine...it's fair for that...but if I want to see an inp of the highest score ever achieved in a game I want to be able to easily find that score....whether it's legit or not or with alphamame or not.

It's a shame as it appears German's Frogger score will be overlooked by most here when if a legit score it's really quite an accomplishment!

Pat was the one that was very suspicious of the recording in the first place. I don't know any specifics as to why, however. But it would appear to have been played at a slower speed, but it's still quite inconclusive as it's quite hard to determine whether or not slowdown was used in a game like Frogger.

LN2 wrote:Yes but the point made above is this Frogger score isn't a new score. It's just trying to reinstate an old MARP score that was removed by someone in the past. A reinstatement of an old score shouldn't be subjected to alphamame blocking cuz it was submitted prior to 3/3/03.

So if you are reaccepting this 600k Frogger score even with the admission of pausing by the player then it shouldn't be subjected to alphamame blocking.

As much as a big issue is made of alphamame blocking here at MARP it must have really been a serious issue at some time for some games. That's a shame.

It became an issue in one of the early tournaments when someone was bright enough to include a screenshot of a game in their zip clearly showing a really low speed.

LN2 wrote:I guess there are many out there that really aren't true gamers. A true gamer would never cheat. True gamers have integrity and respect for the game.

While I agree for some games the speed it emulates at is very important, for many games it isn't.

Frogger isn't exactly a fast paced game. Playing it at even 50% speed really wouldn't make the game hardly any easier IMHO.

You have to be kidding, right?

LN2 wrote:It's all in judging the timing of the moving cars to get across the road and then judging alignment of logs/alligator etc. to get your frog "home".

Timing is always easier to get right when you have twice as long to react.

mahlemiut wrote:Timing is always easier to get right when you have twice as long to react.

If you have played it that way with slower timing a lot then yes...but if you play Frogger at regular speed then suddenly I make you play it in slow motion odds are you are going to have trouble with it.

Think of you standing along side a freeway and wanting to cross it.

Now slow down everything a factor of 3-4 times... is your brain going to know how to time your moves to get across better than versus regular speed? Not likely...not without a lot of practice at that much slower speed.

The other aspect though in mame is when a game isn't running at 100% speed that speed will actually vary some...so if you are using some CPU-hog thing as a "cheat" so the mame emulation is only running at 50% then it's not likely going to be a steady 50%, but varying from 40-60% or more....depending on how the game varies.

That variance in the speed of the game is what screws up timing etc.

I know for newer games in mame that on my older, slower mac can play at only like 80% even using frameskip 6 or 8, it's very borderline unplayable and where I can't play it as well as if I use an older version of macmame that emulates the game much faster so at frameskip 6 or even less I get a full 100% speed.

I always figured that's cuz at 100% speed it's a consistent speed so you can get the timing down. If the speed is varying how on earth are you ever going to get the timing down?


Now for other type games like shooter games like 1942 etc. speed there is very important. If the game is running slower it does make it much easier to see and dodge all the shots...as well as shooting more enemies.

Frogger doesn't have anything like that. If the game was twice as slow as it already is IMHO it would only be very marginally easier....mainly where when a duck appears you have more reaction time to move over on the log etc. so that quick kill doesn't occur. However, for an experienced Frogger player, they already know which "end" of the log to avoid for certain rounds.

Those duck appearances are really about the only "fast" part of the game. The only other odd thing in Frogger is when you have the levels where the cars are slow then they change to fast on you right in the middle of crossing...that part sucks. If the game was even slower then in that case yes you would have more time to react to that change.

Again an expert player knows when those changes are coming though so is careful to not be in the middle of crossing when that happens.

However, my point above was if MARP is reaccepting this Frogger score which seems to be the case since it hasn't been removed again...then you can't subject it to alphamame score blocking cuz this isn't a "new" score.

If you think it was played at a slower than normal speed or the admitted use of pause is no good then remove it yet again! You can't accept it and just block it down as some compromise type of thing.

Those are 2 separate issues. alphamame blocking shouldn't apply in this special case. The pausing and speed issue of this inp is separate from the alphamame blocking issue.

As far as "new" scores being submitted as someone else noted the clear thing to do is in the submit form only list versions that wouldn't be blocked!

I guess this would have to also include 0.36 for use with Sega engine 1 games like Hang-On. No alphamame blocking should apply to those Sega games given no alphamame can properly show the graphics for the game like 0.36 does. If an alphamame 0.36 is made then that blocking would apply for those games also.

It's one of the things keeping me from seriously trying to develop an alphamacmame. I know for macgamers to be happy I would need alphamacmame for the current version, a slightly older version that performs faster, and for 0.36 for those Sega games.

To just have it for the latest and slower version of mame then require all use that slower and in the case of Sega games not 100% working version or have their scores blocked seems unfair IMHO.

If you think it was played at a slower than normal speed or the admitted use of pause is no good then remove it yet again!

None of Negative1's Inps got removed although he admitted using pause as well.

LN2 wrote:

Also, I find it odd just cuz BBH for round 1 of K3 can only get 80% the person managing it just arbitrarily then declares...ok, 80% is good enough speed for this. No exceptions, especially personally based ones, should be made for policies like speed.

1-k3 rules are not the same as regular, only coordinators can decide
2-rd 1 is not a real round, it's just a registration procedure which tends to put more fun than in k2 where all players wanting to enter the tourney had to put their names on the board ; for rd2 and others we will require 95% +

well, perhaps BBH and others are just wasting their time then if some game in a later round requires as much CPU as this first round game does.

I guess personal exceptions will be made again if needed so BBH can participate and potentially advance in the competition.

I can understand decisions if many had that issue...but when only 1 particular player says something and the referee then states...ok, 80% is good for this round so you can play BBH that just doesn't seem right.

Sorry BBH...nothing personal if you are reading this.
I understand the k3 ref can do whatever he wants I guess...but would he have made that 80% exception for round 1 to anyone that had this issue...or is it only cuz it was someone he wants to be in it so gave him special treatment?

I wish I could play...but for one I have a mac and 2 it's certainly at least as slow as BBH's system. I'm sure no exception would be made for me in that case even if I had a similar PC.

EDIT: haha, now I realize you were referring to favoritism seemingly being showed towards me, my reply below had nothing to do with that... oh well, I don't care, I'm leaving it here anyway

no offense taken, LN2. in fact, that's exactly my point for not wanting to submit a recording on 80% (or less) just to qualify... because I don't know what's going to happen in the later rounds. Yeah, the whole surprise element of keeping games secret makes things interesting, but not when there's a good chance of getting a game I have NO way to getting a legal speed on. At least with T6 we knew what the games were beforehand and I could mention anything that I wouldn't be able to run. Imagine getting to the semi-finals (or even the finals, I wish) in this tournament, and then getting something like... oh I dunno, Strikers 1945 III. It would kill me because it's something I *WANT* to play, but I can't legally play it so I lose instantly, making all the time I spent playing the other games wasted.

There was no problem with m35tg3 because games actually ran fast back then and there weren't too many processor-intensive games, but things have changed and some very demanding games have been emulated. I know Phil said on IRC that they were trying to find games that weren't that "big"... it's true Ultimate Tennis is only 1.2 megs or so, not that big at all, but a bigger filesize doesn't necessarily equal a slower speed, especially in this case. So instead I'll be sitting this one out.

One way and perhaps the best way to handle that is if the organizers have selected all of the games for all of the rounds then for that first "entry" round, just make that with the game that has the highest CPU requirement of all of those selected.

Then you know if you can get good enough speed on that particular title...then you are ok for all future rounds.

I would hate to play my butt off to get through the first 4-5 rounds then get down to the final 3 or 6 or whatever players left and withdraw just because a game is then used my system can't play fast enough.

I think the MAME devs are perhaps going too far nowadays to try and get all emulated with one version/package.....at the cost of speed the past several versions.

Perhaps it's time they tried to focus on optimizing the game CPU emulations more now. All the 68k CPU games especially should emulate quite well but seem not to.

Adding more drivers to MAME does not slow it down, bar using a little bit more RAM (the executable does now exceed 17MB).

Other than the ones with high MHz (ie: KI, Cruisin' and a few others), CPU emulation isn't so much the issue anymore. It's just a small piece of it. More powerful graphics and sound hardware add to it all the more. Just you wait until games with REAL 3D hardware start getting added. (Aaron is, afterall, working on 3D/FX emulation for MAME) Or ST-V for that matter. Although I'm quite surprised by the speed of System32 games.

Barry, well, I can't say really for PC mames but for macmame 0.66 is just too slow for my mac.

I can run an older version and have it run certain game CPUs 5-10 times faster than 0.66 does.

Older versions had a lot of PPC assembly optimizations that especially in the case of 68k games was very efficient since emulation of 68k CPU on a Mac is very simple.

However, I think the more recent version(s) must have dropped the assembly optimizations or something.

I can't even play regular pacman at 100% speed in 0.66 without using frameskip. All games I have played with 0.66 I am using a frameskip of 6 or 8.

That's sad...and why I still use version 0.60 for all games except those not supported by 0.60.

My Mac isn't exactly fast by today's standards but I can remember with this same mac once being able to get 500-600fps in pacman with throttle off.

Ok fine...sacrifice some speed for compatibility of multiple games but I think it's going too far now when a 450MHz CPU can't even play pacman at speed without using frameskip in 0.66.

The speed thingie really sucks

There must be some simple explanation for the extreme slowness between MacMAME 0.60 and 0.66. MAME has not slowed down that much between then. Not even the Neo Geo driver. I don't know the make up of MacMAME so I can't really comment. Someone needs to produce an X server for MacOS X that has decent optimisations in it, so that XMAME can be effective. OpenGL build, even.

mahlemiut wrote:There must be some simple explanation for the extreme slowness between MacMAME 0.60 and 0.66.

Well, without looking at the code for both not sure. I'm guessing 0.60 still had at least some assembly for the game CPU pacman and others use....but 0.66 perhaps doesn't.

In 0.60 for pacman with throttle off and sound on at max quality, video double size without scanlines I get about 180fps. In 0.66 with same settings I get about 45-55fps. That's a factor of 3 slower. Go back to a version like 0.37, I got like at least 450-500+ fps. Next time I reboot back to OS 9(likely soon to optimize my X volume), I'll test pac and a few other games in 0.37 also for comparison. Now for other game CPUs...like that used for Punchout, the speed of 0.66 is exactly the same as 0.60. In fact in 0.66 I get higher frame rates on Punchout now than I do on pacman. I get 75fps for Punchout with those same settings in both 0.60 and 0.66...no difference at all. I generally haven't played Neo-Geo games in macmame except a couple so can't really comment there.

Someone needs to produce an X server for MacOS X that has decent optimisations in it, so that XMAME can be effective.

nah, building macmame with absolutely no assembly code so just the xmame c-cores and then just the mac GUI stuff as the front end it runs much slower than macmame currently does. I almost wonder if that's about all 0.66 is. I would need to study the code to see. A couple had tried that type of a c/c++ core only build and only got like 25fps for pacman on a mac faster than mine.

OpenGL build, even.

Hard to say there...if accelerated 2-d blitting perhaps...however, all previous versions of macmame actually had RAVE, OpenGL and even Glide 3d api support but games would either run the same speed using those or slower. For OpenGL, it was always drastically slower. RAVE and Glide were much faster than OpenGL but still slower than software rendering for most games. This was back when I was still using OS 9.

For OS X there is only an opengl rendering plugin but it's not optimized for the latest builds of X and opengl or for the latest macmame so there are issues there. The next build of macmame is slated to be for OS X only. Versions 0.53 through 0.66 were carbon builds so they ran in both OS X and 9. Perhaps with an X-only build some other optimizations for compiling can be used that previously couldn't....or code improvements.

Yes, for specific games using a 2d/3d api can perhaps improve performance...but for older games like a simple pacman game I don't see an API helping much. It's just easier to use 0.60 for that.

I only play 0.60 in software rendering mode. I guess in PC terms that would mean not using direct3d or opengl etc. for rendering.

Yes, I know macmame is slower now cuz I can run mame66 in VirtualPC and for pacman get 25fps(but with no sound) with frameskip 0 and throttle off...so it's definitely more efficient than macmame is now.

Go back to older versions though I think macmame was generally faster for many game CPUs than pc mames were.

Double size? MAME's software stretching is excruciatingly slow. That's the main advantage of using hardware acceleration - stretching is done in hardware for virtually zero performance loss. X11 (Xv), xgl (OpenGL), and xfx (3D/FX) builds of XMAME should provide this, although I would strongly doubt that MacOS X supports Xv, for now at least.

The only ASM code in XMAME relates to the 68000 CPU core, and the dynamic recompiling MIPS cores. This is very little speed difference between the C and ASM 68000 cores, not to mention that the C core is more accurate. Of course, the dynarec MIPS core is more than twice as fast as it is interpreted, but this only affects two games (KI and KI2). You still need around 1.4GHz to run them at full speed anyway.

Under Linux, no hardware acceleration is definitely faster, unless software scaling is used, in which case speed takes a huge hit. DGA2 works best generally, but you have to set up video modes (modelines) in your XF86Config (or XF86Config-4 for XFree86 4.x). I don't know what kind of X server is included or can be used with MacOS X, but I would hope that something at least similar to XFree86 4 is used. I don't know if there is any chance of DGA support in MacOS X, it might be just a Linux thing. DGA2 under Linux gives pretty much the same speed as the Win32 and DOS versions do. But even so, I can still run games upto the speed of SSV games and Neo Geo games at fskip 0 with sound while hardware stretching is enabled. Otherwise, under DGA2, I can run most Konami GX games at fskip 0 (some variances at times though).

One thing I'm curious about though, just how fast do PowerPC CPUs go these days? I've seen recent adverts for PowerBooks at just 1GHz.

Enough banter, people will start thinking that I'm you.

mahlemiut wrote:Double size? MAME's software stretching is excruciatingly slow.

Oh well...again a difference between macmame and pc mames. I am not totally sure but think macmame uses Quicktime for doubling or tripling the size in "software" rendering mode. Yes, it's a tad slower but not really that much. You can add scanline effect so the doubling effect doesn't result in any additional pixels needed. I'll have to do a few runs changing that around also to show you the results.

That's the main advantage of using hardware acceleration - stretching is done in hardware for virtually zero performance loss. X11 (Xv), xgl (OpenGL).

Yes, to stretch noninteger amounts so it fits on any screen resolution.

Macmame provides that also using RAVE, Glide or OpenGL if playing in OS 9 and with OpenGL in OS X.

The only advantage I ever really saw from using the video APIs in macmame was the scaling to the screen size and for some games getting the 3d-smoothing/blurring effect which made some games like marble madness look a tad cooler.

One thing I'm curious about though, just how fast do PowerPC CPUs go these days? I've seen recent adverts for PowerBooks at just 1GHz.

Yes...portables are around 1 GHz...might have one out there that's 1.2 or 1.33 GHz now.

Towers go up to 1.43 Ghz G4 CPUs. The big thing with newer towers really are the dual processors...although macmame isn't written to take advantage of dual CPUs.

This mac is still currently fast enough for most of what I do. My next PC purchase will likely be a PC that is something like at least a Athlon 2200+ or something similar that I could get for around $500-700. I'm not going to go overboard cuz I really would only use the PC for gaming. I would still use my Mac for everything else.

The new CPU for Macs is due out anyday now....supposed to have a demonstration at WWDC in June with models hopefully going on the market later this summer or fall. There are a lot of rumors flying around about that so hard to say.

Anyway, those new CPUs are much faster where the "low-end" will be equivalent to the current high-end Mac.

I'm not sure how fast they have the early ones going now... Rumors vary a lot on that...Last I read there was a 2.5 GHz one but they were still having heat issues with it.

To compare I would need to find someone with a 1.43Ghz G4 and get them to try a few macmame benchmark tests to compare to the speed of my system and speed of mame on PCs.

Luckily for the most part in mame I like the really older school games which are fine in macmame with my system. I'm not into most of the newer Neo Geo games or other games that were out after about 1995.
I haven't even really tried that many of them. I guess cuz I just know I likely would need to use frameskip 8, 9, or 10 to get acceptable speeds...but at that high frameskip control response sucks.