Page 2 of 3

Posted: Sun Apr 07, 2002 4:50 pm
by mahlemiut
MAME in no way manages random seeds, the games themselves do that. And how it does it is likely to vary from game to game.

As for Ninja Commando, I have had no problems playing back Gemant's recordings. Often problems with Neo Geo games are NVRAM related, and I'm pretty sure the Samurai Shodown series works too. (Go Poppy!!! :D)

Posted: Sun Apr 07, 2002 5:02 pm
by Chad
smame wrote:Following games have RTC problem

Metal Slug series
KOF series
Samurai Showdown series
Quiz KOF
Ninja Commando
(The Last Blade/2 ?)
(Twinkle Star Sprites ?)


Following games don't have any problems

Fatal Fury series
Garou MOW


I don't know other games. (no one played others at kaillera)
there are plenty of rbff problems, and a few mslug probs with early mame versions, but all the others you mention playback fine except last blade 2 + kof99 + kof2000. as barry said, have you tried removing the nvram directory? if you post specific URLS to games you can't get played back maybe we can help to play them back for you?

Posted: Sun Apr 07, 2002 5:43 pm
by Kale
It is so difficult? Modify your PC clock to the "last modify" time of the inp file and see if playback...

Posted: Sun Apr 07, 2002 5:54 pm
by smame
Damn, why do you think I haven't removed nvram? You have to learn who is smame! :twisted:

BTW, I remember few games I mentioned don't relate RTC, but mame32 has problems.

I put special mame for marp. Even if NEOGEO feature does not affect, it is useful for marp. Supposing you are interested in it, go to following url.
http://stretchmame.mameworld.net/testver.html

Posted: Sun Apr 07, 2002 7:33 pm
by Chad
FEI: don't click on that link unless you want popups all over the place and want to download a pentium optimized version of stech mame (many are amd users).

sorry about doubting your nvram ability: i wasn't speaking of a mame that might fix problems, i was talking about give us examples of inps that don't work and we'll see if it's a strechmame problem or an official mame problem.

Posted: Sun Apr 07, 2002 7:37 pm
by Chad
Kale wrote:It is so difficult? Modify your PC clock to the "last modify" time of the inp file and see if playback...
the time needed for seeding is the time at the begining of the recording not the end of it. thus unless you know what time the begining of the recording was this would be useless, also it's pretty much of a pain to change the clock of your computer unless you have linux of course and settime.

Posted: Sun Apr 07, 2002 9:44 pm
by mahlemiut
Windows, at least, keeps a creation timestamp. Maybe a modification could be made to MAME that would make use of that when playing back Neo Geo games.

And yes, smame, I do know who you are (well, not your real name or anything, but you know what I mean :) ), and I don't doubt your knowledge of MAME either.

Posted: Mon Apr 08, 2002 1:08 am
by Chad
no dice, linux inps in a tar file would work, windows may have a creation date but the zip file only stores one date which is the last modification date when puting it into zip. so if you zip an inp which all marp recs are, you lose the creation date.

and i dont' think win98 makes these dates right, i'm looking at some inp files that i created a while back, and they have a created date later than the modify date.

created: march 30th, 2002
modified: december 5th, 2001

lets change marp to only accept .inp.tar.bz2 files :)

Posted: Mon Apr 08, 2002 2:50 am
by smame
Chad wrote:FEI: don't click on that link unless you want popups all over the place and want to download a pentium optimized version of stech mame (many are amd users).

sorry about doubting your nvram ability: i wasn't speaking of a mame that might fix problems, i was talking about give us examples of inps that don't work and we'll see if it's a strechmame problem or an official mame problem.
Who still uses K6/II/III? Do you know PentiumPro optimized is the best for Athlon and Duron? Of course PentiumPro/2/3 too.

I fixed RTC problem one year ago(2001-Apr-27). So I don't have any inps that I can't playback.

Posted: Mon Apr 08, 2002 8:26 am
by MKL
Chad, would you please point out to me a good mame version where *Metal Slug 1* has no playback problems?

I tried 057 and 0.36 final and neither seems to playback correctly...

Thanks.

Posted: Mon Apr 08, 2002 12:16 pm
by Kale
MKL wrote:Chad, would you please point out to me a good mame version where *Metal Slug 1* has no playback problems?
Try .34b2(the first version where mslug appared).

About the RTC,pick up the last modify date,check how much long is it(with an hex editor):
every start of a non zero hex digit is a start of a new frame;
every 60 frames (in Neogeo)is a second.
So,in mslug the first frame is 0x21-0x49=0x28(=40dec)
Divide this value for the total number of hex digits and you'll have the total number(Obviusly,don't count the first 0x20 digits!They are information data for MAME i.e the zipname )


By the way,I imagine that the RTC affects only with hours.minutes.seconds,not also with day number.
Question:does WWF Wrestlemania have the same problems?There is a RTC clock also there...

Posted: Mon Apr 08, 2002 1:20 pm
by Chad
gosh, sorry i thought there were some confirmed mslug recordings since it was in a tournament and a lot of recordings were accepted there. I would have suggested m35tg3 to use (since i remember myself playing back a couple of mslug recordings in the tournament) but i can't get bbh's to playback now.

Kale:
even if you assume that the recorder is getting exactly 60 frames per second (which is never the case even with very fast computers), there is still some latency time closing and opening the file that would mess up a creation time estimation from the modification time. There is also the timing issue involved of setting your system clock exactally X seconds behind the original creation date, since it takes time to actually load mame and start the playback. I think it would be more frustrating to write a program that did this to find out it still doesn't playback because the creation time might be off by 10 seconds, and you try moding the time by +1,-1,+2,-2,+3,-3,etc... and if you still can't get it to playback it would suck.

Fixing mame to do the right thing (store or ignore the RTC for recording+playbacking) is the thing to do. But then again i don't like the idea of using an unofficial version of mame to do this: adding yet another thing for a user to search for in the future to playback a recording. it's hard enough to get old mame32 versions, who has strechmame version 37b7?

Posted: Mon Apr 08, 2002 2:14 pm
by MKL
Kale wrote:
MKL wrote:Chad, would you please point out to me a good mame version where *Metal Slug 1* has no playback problems?
Try .34b2(the first version where mslug appared).
Thanks, but the rom has changed since and I would like to find a mame version that works with the updated Metal Slug rom...

Any other suggestions?

Posted: Tue Apr 09, 2002 10:20 am
by MKL
A new Metal Slug inp has been submitted by J.Faria (mame 059)... followed author's instructions but playback does not work for me... however it did for a guy I know. Well, I hope someone else is also gonna try this...

Posted: Tue Apr 09, 2002 1:59 pm
by Kale
Chad wrote: Kale:
even if you assume that the recorder is getting exactly 60 frames per second (which is never the case even with very fast computers), there is still some latency time closing and opening the file that would mess up a creation time estimation from the modification time. There is also the timing issue involved of setting your system clock exactally X seconds behind the original creation date, since it takes time to actually load mame and start the playback. I think it would be more frustrating to write a program that did this to find out it still doesn't playback because the creation time might be off by 10 seconds, and you try moding the time by +1,-1,+2,-2,+3,-3,etc... and if you still can't get it to playback it would suck.
Was just an assumption.Obviusly nobody here is interested to get crazy just to playback and sync the clock...
Chad wrote:Fixing mame to do the right thing (store or ignore the RTC for recording+playbacking) is the thing to do. But then again i don't like the idea of using an unofficial version of mame to do this: adding yet another thing for a user to search for in the future to playback a recording. it's hard enough to get old mame32 versions, who has strechmame version 37b7?
You want a fix that ignores the RTC it during record/playback?Just sent against mame.Here's what to do if you want to use it in a personal compile:


/*In src/machine/neogeo.c*/
/*Add this at line 9*/
extern void *record;
extern void *playback;

/*Change this ...*/
pd4990a.seconds = ((today->tm_sec/10)<<4) + (today->tm_sec%10);
pd4990a.minutes = ((today->tm_min/10)<<4) + (today->tm_min%10); pd4990a.hours = ((today->tm_hour/10)<<4) + (today->tm_hour%10);
pd4990a.days = ((today->tm_mday/10)<<4) + (today->tm_mday%10);
pd4990a.month = (today->tm_mon + 1);
pd4990a.year = (((today->tm_year%100)/10)<<4) + (today->tm_year%10);
pd4990a.weekday = today->tm_wday;

/*...Into this*/
/* Disable Real Time Clock if the user selects to record or playback an .inp file */
/* This is needed in order to playback several games,as they are using the RTC of */
/* the NEC pd4990a as pseudo-random number generator -kal 04/08/02*/
if( record != 0 || playback != 0 )
{
pd4990a.seconds = 0;
pd4990a.minutes = 0;
pd4990a.hours = 0;
pd4990a.days = 0;
pd4990a.month = 0;
pd4990a.year = 0;
pd4990a.weekday = 0;
}
else
{
pd4990a.seconds = ((today->tm_sec/10)<<4) + (today->tm_sec%10);
pd4990a.minutes = ((today->tm_min/10)<<4) + (today->tm_min%10);
pd4990a.hours = ((today->tm_hour/10)<<4) + (today->tm_hour%10);
pd4990a.days = ((today->tm_mday/10)<<4) + (today->tm_mday%10);
pd4990a.month = (today->tm_mon + 1);
pd4990a.year = (((today->tm_year%100)/10)<<4) + (today->tm_year%10);
pd4990a.weekday = today->tm_wday;
}


Hoping that will be added into next MAME...