MAME 0.71 out

Post by **mahlemiut** » Fri Jul 04, 2003 11:01 pm

http://www.mame.net/downmain.html

Have fun with Starsweep, the first playable PSX-based game in MAME.

How playable it is, we'll soon see.

Mr. Kelly R. Flewin · Post by **Mr. Kelly R. Flewin** » Sat Jul 05, 2003 3:48 am

mahlemiut wrote:http://www.mame.net/downmain.html

Have fun with Starsweep, the first playable PSX-based game in MAME.

How playable it is, we'll soon see.

I'm still wondering about the "Borg assimilates Zinc" thing.. makes me expect to suddenly see Rival Schools or Tetris Grandmaster working [drools]

Kelly

Post by **mahlemiut** » Sat Jul 05, 2003 4:33 am

after some 'testing'...

Starsweep is not playable for me at a reasonable speed.

80% at fskip 11, ~60% at fskip 0.

Tetris the Grandmaster is a possibly I guess, but don't hold your breath for Rival Schools just yet. I've tried Tekken and Soul Edge, and neither of them even start up. The Capcom games also have an encrypted BIOS that needs to be cracked too.

LN2 · Post by **LN2** » Sat Jul 05, 2003 6:09 am

mahlemiut wrote:after some 'testing'...

Starsweep is not playable for me at a reasonable speed.

80% at fskip 11, ~60% at fskip 0.

is that supposed to be 6%? cuz 60% at fs0 and 80% at fs11 doesn't jive.

if you are getting 60% at fs0 then you would be at 100% at fs6 or more.

Post by **mahlemiut** » Sat Jul 05, 2003 7:23 am

Nope, that's right. Some arcade systems have higher CPU power than others, thus frameskip can't help that. It's much more noticable on hardware like KI or Cruisin' USA, where the CPU is around 100MHz, and has to do 3D calculations itself.

Haze · Post by **Haze** » Sat Jul 05, 2003 10:01 am

the video emulation uses next to 0 cpu time compared to the cpu emulation therefore skipping frames doesn't help at all.

its just 100% on an xp1800, below that you'll probably struggle

don't hold your breath for the capcom hardware zn games, the method zinc uses to get them working is 100% unsuitable for mame (think list of hacks longer than your average encyclopedia) and proper emulation isn't possible at the moment.

Post by **mahlemiut** » Sat Jul 05, 2003 11:02 am

How about Taito FX-1a/b?

Would be nice to see the Beastorizer bootleg working, even if I have no hope of playing it at a decent rate. Even Zinc struggles with it (under Windows at least, it's twice as fast under Linux for some inexplicable reason)

Haze · Post by **Haze** » Sat Jul 05, 2003 11:47 am

i'm not sure about the taito ones , sorry. the remaining namco ones are the most realisitc at the moment imo.

LN2 · Post by **LN2** » Sat Jul 05, 2003 2:54 pm

oh, well I guess with my slower older system I don't evne try any of those types of games.

I have had some though that required using frameskip 10 to get 100% speed though...and the speed still seemed to scale versus the frameskip.

So with frameskip the only difference is the video drawn? Is the game actually still calculating the input and scene etc. at it's full frame rate?

It doesn't seem to jive some...for example...in older versions of macmame, I could get 400+fps for pac-man. Now with 0.66 I get only about 58fps(sound on).

You would think the video requirements for pac-man wouldn't have changed that factor of 8 in those different versions of macmame. The difference must be in the CPU load from the emulated game CPU itself.

Yet even with a factor of 8 change....the speed versus frameskip scales extremely nicely. ie. if I use frameskip 6 I am up around twice the frame rate(throttle off of course). Using your logic this implies it's all just the video slowing it down.

Haze · Post by **Haze** » Sat Jul 05, 2003 4:27 pm

yes, frameskipping only causes the code in Video Update to be skipped each frame, obviously the game has to run all its code otherwise it just plain wouldn't work.

for some drivers the video update does nothing but copy a bitmap generated by the game code to the screen, skipping it in those cases won't help, in other cases the cpu emulation is so demanding video is insignificant, again skipping won't help.

actually because of the way frameskipping works its important not to put any code that needs to be run every frame in the video update, some old drivers handled collision in video update which causes serious problems if you skip frames.

pacman could have slowed down for several reasons, depending on how old the versions you're talking about are. improvements to various parts of mame can slow things down as a whole, older versions had more video optimizations, newer versions are cleaner and concentrate on just the emulation and easy to understand code.

very old versions would have dirty rectangles whereby only areas of the screen that change get redrawn every frame, they also pre-rotated the tile graphics etc. new versions just redraw the whole screen, and rotate the image during blitting which could be slower on the mac instead but makes for easier to read drivers.

MooglyGuy · Post by **MooglyGuy** » Sun Jul 06, 2003 12:04 am

On the subject of star sweep, once it's on the main page, look who has the #1 score for 1P Challenge.

Speaking of which, there's going to have to be the same sort of split-up that Puzzle Bobble 2, Puzzle Bobble 3, Puzzle Bobble 4, and others have, since there's also a 1P Vs. CPU mode.

LN2 · Post by **LN2** » Sun Jul 06, 2003 5:28 am

Haze wrote:whereby only areas of the screen that change get redrawn every frame, they also pre-rotated the tile graphics etc. new versions just redraw the whole screen,

oh, well if that is what the pc-mame versions are doing now I guess that's part of the difference...cuz macmame still only redraws what changed since the last frame.....at least it seems that way...cuz if I shell out where part of the video ois covered up then switch back, that part is "black" and you can see what's drawn is just stuff that has changed....the static part...like in pac-man the maze part...isn't redrawn until the start of the next board....when the whole screen is redrawn at the start of a new board.

I guess that also explains why for pc-mames you get significantly different frame rates if you use 1x1 or 2x2 or 3x3 sized video for the game. In macmame I see very little difference in frame rates of the game given normal, double, or triple size. This follows the logic that macmame only draws what changes so then the extra requirement from doubling or tripling is quite small..since not much changes each frame in pac-man at least.

I would need to study the code a lot closer to get any more detailed than that.

forums.marpirc.net

MAME 0.71 out

MAME 0.71 out

Re: MAME 0.71 out