TGFC Lifestyle - Powered by Discuz! Board

标题: [老游杂谈] cps1系列至今未被精确模拟？！！！ [打印本页]

作者: lovecf4444 时间: 2012-6-26 21:03 标题: cps1系列至今未被精确模拟？！！！

CPS1是CAPCOM的一系列经典街机的硬件平台，在该平台上的作品包括街霸2、名将等一系列经典作品。虽然目前模拟界的普遍看法是认为这些作品早已被完美模拟了，可是事实确非如此。
我在WINKAWAKS、MAME、FBA等模拟器平台上模拟过CPS1的游戏的感觉是：虽然在画面上这些模拟器的表现十分完美，但是总是在游戏手感上和实机有些出入，总觉得模拟器上的游戏节奏要快于实机，大家也反映过WINKAWAKS有模拟速度过快的问题，而且大家都比较推荐MAME。不过我用了MAME后感觉虽然是比WINKAWAKS在手感上更接近实机，但也是不完美。于是近日上网查资料，结果发现一些国外网友的资料证实的我的猜想。

目前在MAME平台上对CPS系列的模拟（CPS1 2 3）都有一些不完美的地方，其中一点便是运行速度过快。国外网友贴出了一段关于名将的实机视频以加以证实：
在实机的演示画面当中，名将的队长放电的一瞬间脸的方向应该是朝向右面的，而在MAME模拟器中，他却脸向左。有网友指出造成这种现象的原因是因为MAME根本就没有模拟CPU的等待指令，从而造成画面运行过快，解决方法是将CPU降频，不同的CPS1的游戏降频程度也不尽相同，从96%到74%不等。

也有网友提出了权宜之计：就是采用一款名为CALLUS的模拟器，这款模拟器虽然也没有模拟CPU的等待指令，但是它通过HACK手段，将CPU的频率由原来的10MHZ或是12MHZ降到8MHZ,以实现将近实机的速度。
下面我贴出一段国外网友拍摄的名将实机演示画面，该视频由MAMETEST网站的Smitdogg采集自实机基板：
http://smitdogg.mameworld.info/m ... omm-japan910928.wmv
大家有兴趣的话可以用最新的MAME试试队长在演示画面放电时脸的朝向。也请有条件的朋友证实一下实机中演示画面里队长的脸的朝向.

[ 本帖最后由 lovecf4444 于 2012-6-26 21:16 编辑 ]

作者: lovecf4444 时间: 2012-6-26 21:06

这是实机截图
队长脸朝右

[ 本帖最后由 lovecf4444 于 2012-6-26 21:10 编辑 ]

作者: lovecf4444 时间: 2012-6-26 21:09

这是模拟器截图，队长的脸朝左，我自己用MAME 145U5试过，队长的脸也是朝左，和实机相反。不过换CALLUS的话，就是向右，和实机一样。

[ 本帖最后由 lovecf4444 于 2012-6-26 21:12 编辑 ]

作者: LILIT 时间: 2012-6-26 21:16

额，看起来好高深

作者: 核爆前夜 时间: 2012-6-26 21:44

posted by wap, platform: SonyEricsson (Xperia Play)

听着都感觉很厉害，有点折腾啊。

作者: tooomy 时间: 2012-6-26 21:46

玩的欢乐就行了，太讲究那不和水电、煤电那个笑话没区别了~

作者: chucky 时间: 2012-6-26 21:53

Callus 是不是98年出的那个CPS1的模拟器？

作者: lovecf4444 时间: 2012-6-26 22:04

嗯，就是，是一款挺老的，好象是95年左右出的。

作者: tzenix 时间: 2012-6-26 22:06

也许是作者们觉得这无关紧要吧

作者: alexey 时间: 2012-6-26 22:08

坐等几位技术大牛来回答这个问题

作者: tzenix 时间: 2012-6-26 22:14

话说模拟界到今天也发生了很大变化了
以外都是几个team各自为阵，不开源，各做各的
现在主流模拟器都是玩SVN开发了
要是几个CPS1模拟器的作者能合作该多好，也不至于搞得连rom都不通用

作者: dragonzet 时间: 2012-6-26 22:23

这个 LZ你找个PS版本ISO试试看，PS版应该是完全移植的吧，要是玩起来一样就说明没什么问题

土星不知道有没移植过

作者: lovecf4444 时间: 2012-6-26 22:28

手头没有PS1移植的版本，再说谁知道PS1的模拟器精不精确呢？我手头也没有PS1的实机。

作者: lovecf4444 时间: 2012-6-26 22:41

我发现街霸2 的CPU是在10MHZ运行,而我用的WINKAWAKS的默认CPU 频率是12MHZ，这也许就是WINKAWAKS的手感要比街机快的原因？

作者: tzenix 时间: 2012-6-26 22:50

或许这也是我接受不了IIX的速度的原因？不过，这点速度差别不大吧……

作者: lovecf4444 时间: 2012-6-26 23:15

这点速度差别可大着呢，CPU的运算速度会造成敌人AI的变化，敌人的行为方式也会变化，FTG游戏的话,出招节奏会有变化。

[ 本帖最后由 lovecf4444 于 2012-6-26 23:35 编辑 ]

作者: lovecf4444 时间: 2012-6-26 23:19 标题: 回复 11# 的帖子

是啊，洋人搞模拟器的劲头和搞科研差不多，十几年如一日，死磕我们看来根本不重要的问题。他们那种吹毛求疵的态度和我们动不动就完美模拟的态度相差真的是很大。

作者: Yit 时间: 2012-6-26 23:58

posted by wap, platform: iPhone

引用:

原帖由 @lovecf4444 于 2012-6-26 23:19 发表
是啊，洋人搞模拟器的劲头和搞科研差不多，十几年如一日，死磕我们看来根本不重要的问题。他们那种吹毛求疵的态度和我们动不动就完美模拟的态度相差真的是很大。

洋人做什么事都比tg人认真的多

作者: tzenix 时间: 2012-6-27 00:01

引用:

原帖由 lovecf4444 于 2012-6-26 23:15 发表
这点速度差别可大着呢，CPU的运算速度会造成敌人AI的变化，敌人的行为方式也会变化，FTG游戏的话,出招节奏会有变化。

速度还是匀速的吧，我没感觉到出招节奏的变化

作者: 还珠楼主 时间: 2012-6-27 01:03

引用:

原帖由 lovecf4444 于 2012-6-26 23:19 发表
是啊，洋人搞模拟器的劲头和搞科研差不多，十几年如一日，死磕我们看来根本不重要的问题。他们那种吹毛求疵的态度和我们动不动就完美模拟的态度相差真的是很大。

环境不一样~

国内谁会把大量时间放在这上面，房子，车子，妻子，孩子啥不比这些总要~想要这些就得赚钱，想赚钱就不能把时间放在不能为自己带来回报的事情上~

等中国未来成了欧美这种环境自然也有人愿意耐得下性子搞这些了~至少吃穿不愁的情况下~

作者: SONlC 时间: 2012-6-27 01:39

1.脸朝左和朝右的比较是使用某基板和该基板上dump的rom放在模拟器里执行然后进行比较的吗？如果不是，那不能排除名将有其他未dump版本，或者某个版本是hack版。
2.你说callus玩，脸向右，是和mame同样的rom吗？
3.如果把速度降下来，脸的朝向会变得一样吗？如果不会，结论同第一点。

总的来说，脸的朝向和模拟速度的精确与否我觉得是两个独立的问题。
模拟精准是相对的，不精准是绝对的，实际上现代PC根本不可能模拟精准的垂直同步中断，光这一点就秒杀所有16位以前那些所谓已经完美模拟时序的机器了，不信拿个HazeMD或者Fusion模拟一下MD，然后旁边开台实机就知道了，半分钟内即开始明显看到速度差异。

但是从另一个角度上来说，不精准本身也是一种精准，因为没有两块基板的速度是完全一样的，没有一个晶震速度是绝对一样的。。。

[ 本帖最后由 SONlC 于 2012-6-27 01:43 编辑 ]

作者: yangjuniori 时间: 2012-6-27 02:25

posted by wap, platform: SAMSUNG (Galaxy S II)

sf2CE泥马Demo慢投币开始就快了……

作者: 云里翻 时间: 2012-6-27 07:20

我不相信callus比mame更精确。楼主根据表面现象判断是不正确的。。。

作者: dream123 时间: 2012-6-27 07:35

posted by wap, platform: iOS

iphone有个官方的cps1模拟器，算完美模拟了吗

作者: 快乐猪头 时间: 2012-6-27 07:45

posted by wap, platform: Opera

MAME的CPS1驱动改进确实直到现在还在继续，一直就没有停止过。但是，对于LZ的研究对比，猪头惭愧的表示我还没还有达到这个境界，并进一步无耻的表示目前还不想去达到这个境界......

作者: lovecf4444 时间: 2012-6-27 08:28

因为CPU速度差别造成的节奏不一致在玩其他游戏的时候不是很明显，但是玩FTG的话应该是能感觉出来在连段的时候节奏有区别。我没有实机基板，我只是用CALLUS和MAME145U5试过名将，但是CALLUS用的ROM和现在最新版的MAME用的ROM格式不一样，另外MAME怎么降低被模拟CPU的速度？这我还不会。所以SONIC提出的问题我无法回答。自己测试的结果是用CALLUS95和新版MAME145U5在用名将测试片头演示模式中队长脸的朝向绝对不一样，这一点应该不是模拟晶震速度不一致造成的。

作者: hudihutian 时间: 2012-6-27 10:00

这根本算不上不精确了，这根本就是错误的嘛

作者: 云里翻 时间: 2012-6-27 10:08

mame里按tab

作者: James50 时间: 2012-6-27 10:30

同LS，试过降频90%就正常了，神奇

[ 本帖最后由 James50 于 2012-6-27 10:31 编辑 ]

作者: lovecf4444 时间: 2012-6-27 11:20

请问云里翻，具体用哪一版 MAME可以实现升降MAIN CPU的功能，我在用MAME PLUS 145U5,这个版本的TAB菜单里面没有相关功能。

作者: lovecf4444 时间: 2012-6-27 11:26

还有画面刷新率的问题，59.6和59.1怎么测？FRAPS都是显示整数的，59.1是直接显示59,59.6呢？是直接显示60？因为有人有DV拍过MAME的机台，说街霸（CPS1 CPS3）系列的刷新率都不正常。

[ 本帖最后由 lovecf4444 于 2012-6-27 11:28 编辑 ]

作者: 快乐猪头 时间: 2012-6-27 12:06

引用:

原帖由 lovecf4444 于 2012-6-27 11:20 发表
请问云里翻，具体用哪一版 MAME可以实现升降MAIN CPU的功能，我在用MAME PLUS 145U5,这个版本的TAB菜单里面没有相关功能。

SLIDER CONTROLS项目里面。话说这个也没用过的话，LZ就去研究起59.1帧、59.6帧与60帧的手感差异？

[ 本帖最后由快乐猪头于 2012-6-27 12:09 编辑 ]

作者: lovecf4444 时间: 2012-6-27 12:15

你们到底用的是哪一版的MAME ? MAME版本不一样，菜单也不一样

上网搜索出来老外用DV拍的一段MAME机台的录像。
他用的机台是CRT，是带扫描线的，用DV可以拍摄出VBI的走向，他说黑道往上走不正常。
http://www.youtube.com/watch?v=M ... ure=player_embedded

这个老外拍这段视频的目的就是告诉MAME的开发人员，CPS1 CPS3系列的刷新率不正常。

[ 本帖最后由 lovecf4444 于 2012-6-27 12:26 编辑 ]

作者: 孤高的套套 时间: 2012-6-27 12:24

真怪把频率降下来方向真的反了

作者: lovecf4444 时间: 2012-6-27 12:28

不对吧，频率怎么变成58.61了，不是应该是59.61吗，你自己设置了吗？他们说是降CPU的频率，不是降刷新率。

作者: 坳由根 时间: 2012-6-27 12:39

引用:

原帖由 云里翻 于 2012-6-27 07:20 发表
我不相信callus比mame更精确。楼主根据表面现象判断是不正确的。。。

胡扯，你就没见过callus吧？？

我在tg起码说过5次了，玩cps1游戏必须用callus~

所有机种，想要用模拟器玩，最后的一个选择才是mame~~多功能模拟器只给精确模拟要求低的用户准备的~~~

作者: 孤高的套套 时间: 2012-6-27 12:44

现在正常了

作者: lovecf4444 时间: 2012-6-27 12:50

不降频的话，不光队长脸的朝向是反的，在演示画面忍者出刀的节奏也不一样，不降频的话，忍者出刀的节奏要偏快，大家可以仔细观察一下。

作者: 快乐猪头 时间: 2012-6-27 13:04

引用:

原帖由 坳由根 于 2012-6-27 12:39 发表

胡扯，你就没见过callus吧？？

我在tg起码说过5次了，玩cps1游戏必须用callus~

所有机种，想要用模拟器玩，最后的一个选择才是mame~~多功能模拟器只给精确模拟要求低的用户准备的~~~

兄弟，别用胡扯这种语气好吧。

你的说法听起来就像是打电话必须用诺基亚，但实际上大多数人已经不在乎这点了。

CALLUS仅仅存在了两年时间，而MAME为了CPS1的精确度，迄今为止已经干了14年，实际上在最初的两年就已经基本完美了，到现在还一直在继续完善，MAME的程序员真比Bloodlust落后这么多的话，这些工作做给谁看的？

作者: yangjuniori 时间: 2012-6-27 13:19

posted by wap, platform: SAMSUNG (Galaxy S II)

这段视频是我在广元西路正阳拍的让楚江提交的

DV是59.94 如果比这个低那么就是黑线往上滚如果是60就是微微向下很慢

足以证明以前CPS1 CPS3写60.00是错的

NEOGEO是59.185606滚的最快

大部分人很少用NTSC制的DV

作者: lovecf4444 时间: 2012-6-27 13:23

问题是现在大部分的玩家用的都是液晶显示器，在这种情况下，如何判断59.6和60的刷新率呢？画面在液晶显示器上都是经过逐行处理再显示的，没有那个可以滚动的扫描线可以做参照了。

[ 本帖最后由 lovecf4444 于 2012-6-27 13:25 编辑 ]

作者: lovecf4444 时间: 2012-6-27 13:34

再说根据显示原理，液晶的60帧未必比CRT的59.6快，说不定还要慢。

作者: yangjuniori 时间: 2012-6-27 13:35

液晶确实更难发现一点但是液晶还是有刷新率的尤其在广电这一块 50-60HZ的差异还是会造成帧率不平顺比如一些户外LED 播放广告的他们用PC播视频是PAL50HZ的就会跳几帧

只有MAME会有与刷新率同步这种功能想想也真是专业方便我使用特殊分辨率么？

作者: 云里翻 时间: 2012-6-27 14:25

引用:

原帖由 坳由根 于 2012-6-27 12:39 发表

胡扯，你就没见过callus吧？？

我在tg起码说过5次了，玩cps1游戏必须用callus~

所有机种，想要用模拟器玩，最后的一个选择才是mame~~多功能模拟器只给精确模拟要求低的用户准备的~~~

吓尿了。。。

我确实外行，不懂。但是一个98年老模拟器的精确度和mame比，不说cps1驱动，就是底层m68000和音乐芯片的模拟度上，都应该是mame完胜吧？

至于cps1驱动，也有这么多wip，你觉得callus和mame在一个水平线吗

0.31 [Paul Leaman]

Bugs:

- ffight, area88, cawing: 30 impulses / second autofire doesn't work with game design "flaws". Kale (ID 04012)

WIP:

- 0.145u5: CPS-1 update [Team CPS-1]: Fixed some labels in various sets to match real PCBs, added some documentation notes and confirmed PCB infos. Reordered sf2cejx sets.

- 0.145u4: CPS-1 update [Team CPS-1]: Added some notes and done some documentation fixes. Expanded and updated games table in video\cps1.c. Removed the notes about LW-13.10D ROM in forgottnu and RT_23.13B ROM in wonder3, since current dumps are confirmed as good. Fixed qad labels and confirmed PCB infos.

- 0.144u7: Aaron Giles fixed U.N. Squadron and other CPS1 games with OOB tiles and strange behavior while single frame stepping.

- 0.144u3: CPS-1 update [Team CPS-1]: Added LWCHR.3A, ST22B.1A, STF29.1A, TK22B.1A, VA24B.1A and YI24B.1A PALs. Minor cleanups and fixes.

- 0.144u2: CPS-1 update [Team CPS-1]: Minor cleanups and fixes.

- 0.144u1: CPS-1 update [Team CPS-1]: Added PCB infos to video\cps1.c. Added some documentation notes to various sets. Minor cleanups and fixes.

- 0.142u4: Joe Caporale gave labels to many inputs in CPS1 games.

- 0.141u2: CPS-1 update [Team CPS-1]: Added Tenchi wo Kurau (Japan Resale Ver.). Confirmed B-Board # for chikij, added ROMs positions, PALs and some PCB infos to video\cps1.c. Fixed sf2uk labels and added PCB infos. Identified and documented willowje as an early and genuine USA release, added a note and renamed willowje -> willowo. Verified and fixed XTALs of Japan Resale Versions to work at 12MHz, they all use DASH A-Boards. Minor cleanups and fixes.

- 0.140u2: CPS-1 updates [Team CPS-1]: Confirmed B-Board # for forgottnua, willowje, cawingu, fixed their labels where wrong and added pcb infos to video\cps1.c. Minor cleanups and fixes.

- 0.139u2: CPS-1 update [Team CPS-1]: Minor cleanups and fixes.

- 18th August 2010: Smitdogg - The box of Capcom goodies arrived today from Japan. They are Japanese versions of Knights of the Round, Willow, and King of Dragons. In Willow's case we got it to correct the current set using incorrect roms. But for the other 2, yet again, the program rom labels indicated new versions we don't have dumped and so far look to be matching currently dumped sets, or at least doing some mixmatching with program roms and different sizes on graphics roms. I'm not really sure yet to be honest, they aren't fully dumped yet. You'll see why when you see how many roms they use. I'll finish dumping them all tomorrow.

- 10th August 2010: Smitdogg - We won the Capcom games that I was raising money for a few days ago. They should arrive in about a week. There is other oldschool Capcom news on the way as well.

- 0.139u1: CPS-1 update [Team CPS-1]: Expanded and updated games table in video\cps1.c. Confirmed B-Board # for 3wondersu and varth. Fixed labels in ghouls, ghoulsu to match real PCBs. Removed 14.bin in dynwarj, it was a bad dump. Marked nm_23.12b in nemoj BAD_DUMP. Removed knightsj set (roms were split & hacked to work on a suicided board). Added roms relationship between parent/clones where missing. Fixed some incorrect notes added in previous update. Reordered some sets. Minor cleanups. Fixed sf2ceua labels and added PCB infos. Fixed sf2ceub labels and added PCB infos, removed 1 incorrect rom and replaced with proper redump done by Dumping Union. Fixed punisher and punisheru Q-Sound labels and added D-Board PALs. Brian Troha set correct B-Board mapper for "resale" kits. This fixes missing graphics in Area 88 (Japan Resale). Documentation update for some CPS1 sets (B-Board chip information) verified via Team Japump PCB photos [Brian Troha].

- 1st August 2010: Smitdogg - Team Japump are dumping/fixing some CPS1 revisions. I got our bids in for the other needed Capcom revisions.

- 0.139: CPS-1 update [Team CPS-1]: Added missing PALs as reference to most of remaining sets. Dumped and added TK24B1.1A to Dynasty Wars [Corrado Tomaselli]. Fixed LW-13.10D rom loading in Forgotten Worlds to match real PCB. Identified, verified and documented a ton of PCBs, adding/fixing missing/wrong infos as ROMs labels and positions, B-Board, C-Board and CPSB IDs and PALs [Stefan Lindberg, dlfrsilver, MKL]. Reordered all sets per release date in cps1.c and video\cps1.c. Updated games table in video\cps1.c and added some notes on PALs table. Minor cleanups.

- 0.136u1: Fabio Priuli fixed NVRAM for dino, punisher, wof and clones

- 0.135u4: Fabio Priuli added driver data struct and save states to CPS1 driver.

- 0.133u3: Corrado Tomaselli verified and added correct xtals values for CPS1 boards. Added correct xtal and divisor for OKI6295 taken from Forgotten Worlds schematics.

- 0.129u4: Nicola Salmoria fixed a contact MAMEDEV message in the CPS1 driver.

- 0.128u4: Updated CPS1 documentation [Nicola Salmoria].

- 0.127u1: Additional input port cleanups in the CPS1 driver [Fabio Priuli].

- 0.127: Nicola Salmoria updated CPS1 to use proper gfx ROMs (+ PALs) for Varth (Japan 920714), and to load correct A-Board PALs for QSound games.

- 0.126u5: Corrado Tomaselli and Charles MacDonald added reverse engineered PLD images to CPS1 games (A-board to all and B-board and C-board to many).

- 0.126u4: Corrado Tomaselli verified and corrected CPS1 refresh rate. Changed VSync to 59.610000 Hz.

- 0.125u6: Corrado Tomaselli changed all games after Street Fighter 2 CE to have a 12MHz clock. Verified on Mega Man and Quiz & Dragons (original kits with plastic case). I am resonably sure all the other games after ST2CE have the DASH CPS1 board with 12MHz clock.

- 0.125u3: David Haywood fixed duplicate input port bits in CPS1 driver.

- 0.124u5: Massive update to ROM naming in the CPS1 driver, also loading smaller EPROMs for Japanese sets instead of the larger mask ROMs where possible. Added information about B-Board revisions where possible. There is still work to do on this regard but this is a first step in untangling the mess [Nicola Salmoria]. Nicola Salmoria and Corrado Tomaselli verified more CPS1 gfx ROM mappings against PALs. Fixed rom names in all games.

- 0.124u4: Further CPS1 improvements [Nicola Salmoria, Corrado Tomaselli, Charles MacDonald]: Fixed layer enable at the end of stage 4 in Varth. Increased Varth CPU clock to 12MHz to reduce slowdowns (the game was released after sf2ce so the faster A-board was available). Verified more gfx rom mappings against PAL dumps (thanks to Corrado). Verified memory maps using A-board PAL dumps (thanks to Charles). Separated the address ranges for CPS-A and CPS-B customs. This finally clarifies why some registers are fixed and some change with every game. Identified three new unknown registers in the CPS-B-21 custom. Their purpose is unknown. All unhandled accesses to CPS-A/CPS-B registers will now be reported. There shouldn't be any left. Look out for "contact MAMEDEV" popups. Cleaned memory maps. Moved Forgotten Worlds special input memory handlers to DRIVER_INIT. Moved extra input handlers to video/cps1.c since the extra connectors are on the C-board. Added sound to Final Crash. Lowered SF2 clock speed to 10MHz. This is on the assumption that, since the 12MHz A-Board is called "89626A-4 DASH", it should have started to be used with SF2 DASH--that is, sf2ce.

- 0.124u3: Nicola Salmoria and Corrado Tomaselli fixed CPS1 palette control register to match tests on Magic Sword PCB. Nicola Salmoria verified several CPS1 graphics ROM mappings against PAL dumps. Nicola Salmoria merged CPS1, CPS2 memory maps and some tweaks from schematics, though to get perfect memory maps dumps of the A-board PALs would be needed.

- 0.124u2: Verified some GFX rom mapping tables against PAL dumps. Improved understanding and emulation of palette related registers [Nicola Salmoria, Corrado Tomaselli]. Nicola Salmoria removed CPS1 tile kludges and replaced with per-game ROM mapping tables. Ideally, the ROM mapping should be derived from mapping of the PALs found on the B board. Note that this is a significant change and some new bugs might be caused by it. Unlike previous bugs, where extra garbage tiles would appear in the games, the new bugs would be tiles or sprites MISSING--so they will be harder to notice. This would happen because in some cases the 16x16 regions of the gfx ROMs can be used by both scroll2 and sprites, while in other cases they must be separated.

- 0.124u1: Nicola Salmoria fixed CPS1 background color. This fixes 3 wonder (game start) and mtwins (explosion during attract). It seems to break Varth, however, and has dubious effects on the boot sscreens of avsp, sfa3 and possibly other CPS2 games. Changed palettesize to 3072 colors.

- 0.122u8: David Haywood fixed the CPS driver to at least display the correct clock in the driver info, by degrading the clock through alternate means.

- 0.121: Vas Crabb updated the DIP locations for all games in the CPS-1 driver.

- 0.114u2: Couriersud fixed crash if you attempt to view graphics page 4.

- 0.114: Aaron Giles fixed a MAME crash if you do a hardware reset.

- 0.112u3: MAME Italia forum connected brightness support on the CPS1 board (not just CPS2) after verifying that the real board does support it.

- 0.109: Input port cleanup and additional comments in the CPS1 driver [Stephane Humbert].

- 0.105u1: David Haywood updated CPS-1 driver to more accurately draw tilemaps, based on evidence from a board with mixed ROMs.

- 5th April 2006: David Haywood - A guy called Murray posted on the MAME.net and neo-geo.com forums about a problem he was having with his Final Fight PCB. The problem looked like it might be due to a mix of Japanese and US roms as the Japanese roms don't contain a full Winners Don't Use Drugs Logo. He got a friend to dump the socketed roms on the PCB and sure enough the socketed graphic roms matched the Japanese set in MAME. This set, being a newer US revision was then plugged into MAME (ffightua in .105) but there was something wrong. MAME displayed the Winners Don't Use Drugs screen perfectly with this mixed set of roms. At this point I wondered if maybe one or both of his Mask Roms were also from a Japanese revision so I loaded the US set with an entire set of Japanese GFX. The logo was broken, but not in the same way as the PCB. For a while this was rather confusing, it must be getting the Japanese characters from somewhere, but the only place they exist are in the Japanese roms, and it would appear the text layer graphics aren't used in them when running the US set. No matter what I tried to change in terms of the roms I couldn't reproduce the exact problem shown. Then I remembered, MAME had some code to force the 8x8 text layer characters to always be drawn from the Left Hand Side of each 16x16 tile (apparently for Carrier Air Wing...) but the left and right sides of almost every 16x16 character used for the 8x8 layer were identical. The screenshot made it quite obvious that the real hardware does not always draw the left hand side of the tile, but instead alternates from left side to right side for every column on the display. By implementing this behavior in MAME I was able to reproduce the same glitch shown on the real PCB when mixed roms are used. I suspect the problem with Carrier Air Wing which resulted in the left hand side always being used is another problem, the CPS video hardware is full of strange little hacks for the early CPS games anyway. The change will hopefully be in 105u1, I can't see anything broken by it, but some testing would be helpful.

- 0.101u1: Phil Bennett corrected CPS1 frequencies based on measurements from a real PCB. Changed Z80 CPU2 clock speed to 3579545 Hz and YM2151 to 3579545 Hz. And changed 68000 CPU1 clock speed to 12MHz and Z80 CPU2 to 8MHz in games with Q-Sound.

- 0.94u2: Aaron Giles fixed CPS1 QSound routing.

- 0.93u1: Aaron Giles fixed OKI frequency to 7576Hz in the CPS1 games and in the QSound games properly remove the YM2151.

- 0.93: Changed OKI6295 clock speed to 6061 Hz.

- 0.73: Fixed some dipswitches in all games.

- 0.70u2: Some CPS1 dipswitch / input fixes [El Condor].

- 13th December 2002: Stephane Humbert fixed the inputs in the CPS-1 driver.

- 0.57: Fixed missing 'MAGIC SWORD...OTHER COUNTRY' and 'INSERT COIN' text in Magic Sword and Varth.

- 0.53: Moved drivers\cps1.h to includes\.

- 7th May 2001: Chack'n cleaned up the CPS-1 driver romset definitions a little.

- 21st February 2001: Nicola Salmoria added buffering to the keyboard input code, allowing more than two keypresses at the same time. This fixes Zangief's spin in Street Fighter 2.

- 10th January 2001: Paul Leaman added the necessary modifications to the CPS-1 driver to allow CPS-2 emulation.

- 30th September 2000: Darren Olafson fixed a bug that prevented sf2accp2 from working.

- 0.37b6: Changed palettesize from 2048 to 4096 colors.

- 27th August 2000: Nicola Salmoria perfected the CPS-1 starfield graphics emulation (used in Forgotten World and Strider).

- 21st August 2000: Nicola Salmoria added preliminary starfield emulation to the CPS-1 driver.

- 10th August 2000: Nicola Salmoria fixed the protection bugs in Slam Masters.

- 0.37b5: Removed vidhrdw\cps1draw.c.

- 14th July 2000: Nicola Salmoria fixed some CPS-1 graphics problems.

- 10th July 2000: Nicola Salmoria fixed the protection problems in Slam Masters.

- 5th June 2000: Paul Leaman added Warriors of Fate to the CPS-1 driver.

- 28th December 1999: Paul Leaman removed another kludge from the CPS-1 video driver.

- 27th December 1999: Paul Leaman added protection work-around to Slam Masters and did some other CPS1 fixes.

- 19th November 1999: CAB fixed some sound related things in CPS-1 driver.

- 0.36b12: Changed Z80 CPU2 clock speed to 6MHz.

- 0.36b11: Changed sndhrdw\cpsq.c to sound\qsound.c/h.

- 0.36b8: Bryan McPhail fixed sprite lag in CPS1. Nicola Salmoria added vh_eof_callback to the MachineDriver structure. See cps1 for usage examples.

- 18th October 1999: Bryan McPhail sent in a general sprite lag fix, helping CPS-1 games at least.

- 0.36b1: Added vidhrdw\cps1draw.c. Paul Leaman fixed bad tiles in some CPS1 games.

- 8th July 1999: Paul Leaman sent another CPS1 video driver update with all the garbage hopefully gone.

- 0.35: Added sndhrdw\cpsq.c.

- 3rd July 1999: Paul Leaman fixed several CPS-1 graphics problems such as the missing van in Punisher and some others.

- 30th June 1999: Paul Leaman sent a new CPS-1 driver with stub functions for the Qsound system. It doesn't play anything yet, but simple sample playing shouldn't be far off.

- 0.35RC1: Brian Lewis and Andrea Mazzoleni added new tweaked VGA modes: 384x240 (for CPS1, Pang etc.), 384x256 (for Lode Runner etc.), 336x240 (for Gauntlet and other Atari games), 320x240 (for NeoGeo and others).

- 0.35b13: Removed machine\cps1.c.

- 0.35b4: Added machine\kabuki.c.

- 23rd February 1999: The CPS1 driver should now have mostly correct row scrollings and priorities.

- 0.35b3: CPS1 games are being worked on, they are temporarily slower but accuracy is improved.

- 13th February 1999: More progress on the CPS1 driver. Some of the later CPS1 games utilizing the Q-sound system are now working (such as Cadillacs & Dinosaurs, Punisher and Warriors of Fate), and a few minor bugs have been fixed. No sound though, because the Q-sound roms are encrypted.

- 0.35b2: Sound fade timer ports in the CPS1 games (see Willow's Test mode for an example) [Paul Leaman]. Improved row scrolling in Street Fighter 2, plus several other fixes to the CPS1 games [Paul Leaman].

- 0.33b7: The CPS1 games use the dynamic palette system [Aaron Giles].

- 0.33b2: Paul Leaman improved speed in the CPS1 games.

- 0.31: Added cps1.c/h driver, machine\cps1.c and vidhrdw\cps1.c.

作者: 云里翻 时间: 2012-6-27 14:28

引用:

原帖由 lovecf4444 于 2012-6-27 11:20 发表
请问云里翻，具体用哪一版 MAME可以实现升降MAIN CPU的功能，我在用MAME PLUS 145U5,这个版本的TAB菜单里面没有相关功能。

按tab以后，在参数调整里面，你的mame里面应该有的吧。

你说的老外讨论串在哪儿呢？我也去瞧瞧

作者: lovecf4444 时间: 2012-6-27 14:43

http://www.mametesters.org/view.php?id=408
我用的是MAME PLUS 145 U5,算是比较新的版本了，按TAB是有参数调整的菜单，但是进去没有升降CPU的选项，估计你们用的是比较旧的MAME?

另外贴子里报告的sf2hf，是街霸2的16人版，我试了一下，确实比实机要快。关键是别的CPS1游戏快多少都不知道。

作者: lovecf4444 时间: 2012-6-27 14:48

引用:

原帖由 云里翻 于 2012-6-27 14:25 发表

吓尿了。。。

我确实外行，不懂。但是一个98年老模拟器的精确度和mame比，不说cps1驱动，就是底层m68000和音乐芯片的模拟度上，都应该是mame完胜吧？

至于cps1驱动，也有这么多wip，你觉得callus和mame在 ...

新的不一定就比旧的好，MAME在CPS1的模拟的精确性上我看确实不如CALLUS。

作者: 云里翻 时间: 2012-6-27 15:06

引用:

原帖由 lovecf4444 于 2012-6-27 14:43 发表
 http://www.mametesters.org/view.php?id=408
我用的是MAME PLUS 145 U5,算是比较新的版本了，按TAB是有参数调整的菜单，但是进去没有升降CPU的选项，估计你们用的是比较旧的MAME?

另外贴子里报告的sf2hf，是街霸 ...

你可能要打开cheat模式才能调整cpu频率。

作者: alexey 时间: 2012-6-27 15:34

mark

MAME对我这样的俗人足够了

作者: yangjuniori 时间: 2012-6-27 17:56

posted by wap, platform: SAMSUNG (Galaxy S II)

关键最简单的CPU频率刷新率

作者: amaterasu_gz 时间: 2012-6-27 21:40

其实不仅仅是CPS1，MAME模拟出来很多游戏都偏快。最明显的是Raiden雷电，街机台上子弹一多就慢镜头的感觉在MAME里是完全不存在的。其它我感觉有偏快的除了SF系列外还有Konami的出击兵蜂、南梦的龙魂2以及空牙等等。这些游戏只听音乐就能感觉到明显的偏快。Striker1945系列也快了不止一点半点。

[ 本帖最后由 amaterasu_gz 于 2012-6-27 21:42 编辑 ]

作者: 709394people 时间: 2012-6-27 22:52

引用:

原帖由 amaterasu_gz 于 2012-6-27 21:40 发表
其实不仅仅是CPS1，MAME模拟出来很多游戏都偏快。最明显的是Raiden雷电，街机台上子弹一多就慢镜头的感觉在MAME里是完全不存在的。其它我感觉有偏快的除了SF系列外还有Konami的出击兵蜂、南梦的龙魂2以及空牙等等。这 ...

雷电的盗版机板才有慢镜头，正版跟mame模拟的差不多

[ 本帖最后由 709394people 于 2012-6-27 23:09 编辑 ]

作者: yangjuniori 时间: 2012-6-28 00:02

有能力的可以用示波器什么看看图像刷新率

作者: 春麗 时间: 2012-6-28 15:22

玩街机模拟器，个人只认MAME。暂时的模拟度未完善，凭籍MAMETEAM的执着和技术水准，相信只是时间问题罢了。作为我等骨灰街机仔，能玩出童年时代的怀旧感觉足矣。

作者: yangjuniori 时间: 2012-6-28 16:55

posted by wap, platform: SAMSUNG (Galaxy S II)

MAME已经够好了

作者: tzenix 时间: 2012-6-28 18:00

粗枝大叶一点比较开心。过于追求完美很累很痛苦

作者: 模拟器高手 时间: 2012-6-29 10:46

mame主攻方向是兼容性，并且在这十几年里已经甩开其他模拟器几条街了，要论精度，自我感觉比抽象数字更有意义

作者: 乌鸦 时间: 2012-6-29 15:39

能玩就可以了，这么纠结于细节，你们不觉得已经失去了最初的目的了么

作者: Advancedadvance 时间: 2012-6-29 20:32

双截龙二电梯关、收割机关那里的拖慢也是盗版的缘故吗？用mame玩也拖慢，要是能正常速度就好了，儿时就觉得不爽

作者: meizizi 时间: 2012-6-30 18:27

服了，你们都是骨灰级

作者: alexey 时间: 2012-6-30 18:42

引用:

原帖由 Advancedadvance 于 2012-6-29 20:32 发表
双截龙二电梯关、收割机关那里的拖慢也是盗版的缘故吗？用mame玩也拖慢，要是能正常速度就好了，儿时就觉得不爽

这个据说是精确模拟原版硬件的结果

作者: 小僵尸 时间: 2012-6-30 18:58

引用:

原帖由 Advancedadvance 于 2012-6-29 20:32 发表
双截龙二电梯关、收割机关那里的拖慢也是盗版的缘故吗？用mame玩也拖慢，要是能正常速度就好了，儿时就觉得不爽

敌人数量超过3个就会感觉拖慢了，在XBOX用的MAME上，使用1G的CPU可以解决这个问题，副作用是整个游戏速度提高了至少百分之三十

欢迎光临 TGFC Lifestyle (http://tgfcer.com/)