Rule 2g, 90% Minimum Speed

[LordGaz]

Can someone please explain the reasoning behind the 90% minimum speed for recording inps? Why is it 90% and not 100% or even 99% to allow for a margin of error?

IMO, a game that is running at 90% of full speed is much easier. I used to play Dangerous Seed on a machine that only managed run it at around 90% speed and I could nearly finish it! but at 100% I can't get anywhere near as far.

Surely by now we all have machines that can run the majority of the games at full speed and things are also more strict now with alphamame enforcement. So whadya say

Gaz.

[The TJT]

Maybe 95?????

If the speed decreases for split second to something low, that shouldn't count though. many games do it, it only makes things harder. But I agree average % should be higher than 90. If your speed constantly stays at 90..that could really help...If I get 10% more hits at TF 100m....Huh!

[mahlemiut]

Surely by now we all have machines that can run the majority of the games at full speed...

BBH is sure gonna bitch about that.

[Haze]

just remember 99% of systems can't run the startup tests on some games at full speed even if they can run the games just fine, thats going to drag down the average regardless of your pc, unless that part isn't counted when you decide if to dq a recording or not. (see all the SH2 based games for example)

[QRS]

Sometimes a slower speed can make the game harder too

[LN2]

Lordgaz, yes, the old arcade games that have been emulated in mame since 0.30 play fine for all....but many newer games added in recent versions of mame won't play at full speed even with a fairly high frameskip even on a decent speed CPU.

I agree though that 95% would be a more fair number.

I have always stated it would be quite easy for someone to have movie encoding going on or mp3 encoding etc. to bog down the CPU more so they could intentionally slow a game down a little bit to make it easier.

QRS, yes, I find for most games if they aren't at 100% so a consistent speed it makes the game harder. I always add frameskip until the game is at 100%...and if I have to go beyond frameskip 8 I won't even play the game cuz control response just isn't good enough at 8 or higher.

However, think of all the games with time limits that are the logic games etc. Those would all benefit from it running at only 90%. That gives you 10% more time to plan each move in the 200+ mahjong variants in mame. For those quiz question ones it makes that bonus points count down 10% slower so your replies are in a sense 10% faster so more points, etc.

I would be in favor of making that 95%. I think 95% is what TG requires...but would need to check. I think that might be worthy of a poll.

[Skyline]

A few ideas/questions...

a.)Many games that have recently become playable, and/or those who have benefited from a particular bugfix may still have poor performance on a sub-par machine, or even what Micro$oft and Intel consider "advanced" nowadays.. The bootup process of for example some of the Kaneko SuperNOVA and SH-2 games taxes processing power, as it required a lot to run in the first place. Seeing as how startups dent the average play speed in some cases, can't the average speed start from a set point? For example, when the attract mode stars/a 'coin' is inserted? Dunno if that's possible or not.. That way the actual in-game speed will be measured, where the player interacts with the game, not sitting there picking his nose waiting for EEPROM tests to run.

b.) Has there been any mass use of "intentional slowdown" to bring up this case? I ask this because a whole buttload of the games in MAME run slower than the actual machines anyways...regardless of what system you're using. Some of it's MAME version specific, some of it's game/driver specific...You might have an extra second of lag or a few frames lost here and there but it'll still run at a respectable speed.

c.) Can background programs/methods of using intentional slowdown be detected by AlphaMAME?

As for the 95% for all recs...eh. Not that much of a difference for your average gamer between 90% and 95%...the rule of thumb would be to attempt all recs. at 100% right?
Don't rush a good thing. We already got AlphaMAME for cheat prevention, which we know needs that little extra processor percentage to run. Let's not "add salt to the wound" so to speak...

Heh..stupid question but what the heck. How many here would actually spend upwards of $1,000+ to get a system just for submitting to MARP? o.O

[LN2]

As for the 95% for all recs...eh. Not that much of a difference for your average gamer between 90% and 95%...the rule of thumb would be to attempt all recs. at 100% right?

Well, you will find some games really struggle to run at a pure 100%. I see many that even if supposed to be 60fps for some reason hang at 58-59fps instead of 60. 58/60 is down at 96.7%. It's not CPU there cuz I can turn off the speed throttle and get 150-200+fps for the game...so it's some flaw in the game cpu emulation or the rom set that results in that.

Because of that you really can't set a 99% rule or 100% rule. 95% covers those cases though. I haven't seen a game that unthrottled runs well beyond 60fps that when throttled runs at less than 58fps...unless designed that way...like star wars, which really is a 30fps game..but then you see 30/30 in the stats.

Don't rush a good thing. We already got AlphaMAME for cheat prevention, which we know needs that little extra processor percentage to run. Let's not "add salt to the wound" so to speak...

Exactly, but if there was an alphamame of let's say version 0.53 or 0.58, for many games that would run faster than regular mame62 and later so even with the encryption stuff you are gaining from being able to use an older version to play.

Those of us that have played games or watched to confirm under older versions know how much faster mame53 is than mame62 or later for many games. I'm always a little confused why the newer mame core versions don't have conditionals to split itself so for certain games it uses this for performance and for others uses the new code for compatibility instead of it just using the compatible code for all which results in slowdowns for many of the games that don't need that running in compatible mode.

[piot]

Hi all.

just my 2 cents...

1) Many friends of mine only have CPU @ 500Mhz or less. Those guys have many problems to play between 90% and 100% unless playing without sound (not a good solution ).

2) Some games are naturally slower than what they should be. For example I played last week on Street Hoop / Street Slam / Dunk Dream. Even if my computer is not too old (Athlon 1900+) the framerate is near to 80%-85% (with auto frameskip) when I choose the team and CPU has more power than needed. I guess there many time during a game where the FR decreases like this without any reason.

[LN2]

I'm curious there...try that game again but without autoframeskipping on.
Then manually adjust the frameskip first pressing F11 to see the stats while a game is running then using F8 and F9 to change the skip.

I'd bet you get 100% then. Autoframeskip theoretically should be auto-increasing the frameskip until the game runs at 100%...so it seems odd you are getting only 85% with that on...think that autoframeskip has some flaws to it.

I never use it myself...and when playing a game in mame the first time I just F11 then manually adjust frameskip if necessary.

As far as a 500MHz CPU being too slow for some games...yes of course. I am in that boat myself.

There are 3000+ games though so if you can't play 300-500 or so of them cuz of performance big deal...play the other remaining 2500. That will keep you busy a while. Most that have the performance issues are NeoGeo games anyway.

[QRS]

I don´t agree with the Neogeo part Rick. I could run all the neogeo games on my old pc (p2 300) sound on and frameskip 0. 100% speed. I could even run them without sound on my mothers p1 166

The worst games are Cruizing USA, Stun runner and games like that. No way I could run those.. even without sound

Mame has been slower too.. I think that BBH could run the neo geo games on his pc 1 year ago, but on the recent versions he can´t.

Maybe it is different when using a Mac? I mean that the neo geo runs slower on a mac or something? (Just guessing here)

Cheers!

[LN2]

I don´t agree with the Neogeo part Rick. I could run all the neogeo games on my old pc (p2 300) sound on and frameskip 0. 100% speed. I could even run them without sound on my mothers p1 166

Yes, with older versions of mame. Have you tried those neogeo games with modern versions of mame? They run significantly slower than they did back in versions like 0.53.


...at least in macmame that's the case.

[LordGaz]

I have always stated it would be quite easy for someone to have movie encoding going on or mp3 encoding etc. to bog down the CPU more so they could intentionally slow a game down a little bit to make it easier.

Of course now there is an option in MAME to give the emulation a high thread priority so that playing music in the background for example is less likely to cause stuttering during the game.

c.) Can background programs/methods of using intentional slowdown be detected by AlphaMAME?

Yes, that is one of the purposes of alphamame, press F11 during playback to see the recorded speed.

Those of us that have played games or watched to confirm under older versions know how much faster mame53 is than mame62 or later for many games.

This is an unfortunate price of progress. Higher and higher levels of abstraction (eg. DirectX) are being introduced into programming all the time to make producing bug free stable code faster and easier, to hell with efficiency, handcoding assembly language and low level optimization is now a thing of the past .

Nowadays if a program doesn't run fast enough, it's not a case of trying to optimize the code, it's a case of getting a faster processor, sad .

Gaz.

[mahlemiut]

Neogeo games used to work 60fps on my P2 233 with sound until around the early stages of the 0.37 beta cycle, where the 68000 core was switched back to the C core. Then later on, raster effects were correctly emulated, and required the IRQ function to be called for each scanline, rather than just each frame, plus games that abused the raster effects were waaaaay slower (look at World Heroes and Galaxy Fight). I don't know if its still enabled, but F1 could disable rasters for a good speedup on those games. Blame SNK for using hardware that has no rowscroll.

[LordGaz]

Barry, can you explain why Space Invaders when F10 is pressed, runs at 5000-6000% on tgmame35 with sound on but on the latest MAME I can only get around 500% when F10 is pressed whatever settings I try. That's a pretty drastic cut in efficiency I would say. This is on an Athlon 2000XP (1667MHz) CPU.

Gaz.