Arcade Simulator - Layout Permissions

[andrew96]

 

for one location in the lamp table (04) MFME uses  #40  BUT the real chr chip returns #4A in first pass, #26 on second pass and #C2 on  third pass..... THIS is the difference....... when the MFME values are fed to a real machine that's where it can send it paying the wrong percentage, (usually its very tight!)

Definitely something barcrest did with the 3 different figures!!!! but what!!!

 

 

 

 

[serene02]

But which machine specifically has this issue Andy?  Where is the proof of gameplay being affected?

J

[johnparker007]

7 hours ago, andrew96 said:

its difficult to get any info on anything, cant even get a picture of behind a lock showing the position of the switch that mine doesn't have!

Sure thing   But this statement you've made confidently a few times in different ways:

"Just a heads up if using a patched lamp table in the rom its not likely to play correctly in a real machine"

And back in February you posted this one for instance:

"Anyway I am just passing on what has been found out in a real machine with a patched fixed 1 lamp table.  It definitely makes for poor play in some machines. Maybe in some machines its perfect.....but who knows which roms were written with this wider implication in them to prevent CHR hacks!

I agree rom patching is good for MFME as your not playing with real money, its for the fun of it! but in a real machine in a real arcade it is dangerous for sure!!!"

So you have this definite info that a patched lamp table affects payout/gameplay, but no idea what machine(s)?

This could be for instance using one of my early patches, not a new stealth patch... or a patch someone else made along with other hacks causing the issue... there's just so little info to go on   Just that it's definitely 100% really broken, but it's impossible to say the name of a single game it's broken on!  Which does kind get us stuck 

Ah well, I don't have an MPU4 test rig like yours so I've no way to test any of this ideas about broken payout percentages on real hardware... so I'll perhaps come back to this when I've got the time and room to install a real machine/test rig...

If you ever do find any information about this poor play problem you keep posting, ie. just one game it actually happens on, please do update us, as then I can get to the bottom of it!    

Edited May 16, 2022 by johnparker007

[andrew96]

ok well take it then that everything is good and fine.

this is what I am up against, he last looked at messages 10:55PM but mine I sent this morning remains delivered but not opened!

[andrew96]

I won't mention it again unless he gives me info on the machines

 

It doesn't directly affect me anyway, My machine has the proper CHR chip and I have done my investigation into CHR's as far as I am going to go, the test rig is about to be stored away now as I need the bench space!

cheers

Edited May 16, 2022 by andrew96

[andrew96]

[serene02]

Depends what version of cloud 9?  The machine would play fairly tight as it saves up for a huge streak, especially for back then. 

Who is it Andy? Someone over at the Mecca? Do you know who hacked the roms at all?  Probably not one that JP has done.

 

 

J

[johnparker007]

Thanks very much for this info @andrew96, that's super helpful  

I've developed a technique earlier tonight after thinking about this potential Chr issue that can quickly evaluate long term payout % from a cleared RAM, just with this in mind... so I can now test a unpatched ROM in Cloud 999, then a patched one - and see how the payout % differs between the two, over say £5,000 in each ROM.

Hopefully we'll get to the bottom of it   

Fast playing machine demo:

 

[johnparker007]

God damn I've got sucked in now, I'm going to be tired tomorrow!  

Ok so a preliminary test (on Barcrest Cloud 999) - used this layout, set at 88% payout:



Cleared RAM, after ~£300 original rom seemed near target %age:


Chr-free patched version of above rom is actually significantly higher:


I'll need to give these both a proper run tomorrow for more hours in case that's just variance, and it needs more chucking through... does seem higher on the patched ROM, from this quick test though 

Original and patched ROMs:
clnv.zip

So the idea of the 'stealth patches', or at least some more investigation will probably need looking into, unless they settle to the same %age over more play.  £300 is quite a chunk of change though...

Edited May 17, 2022 by johnparker007

[serene02]

But that’s not playing tight as reported?

How odd?  It could of course be variance, it’s a lo tech, they didn’t drop many wins straight in so could be randomness of auto play?

J

[serene02]

Incidentally Cloud 999 will be very generous after a ram reset?  Even so 98 after 300 in does seem very fair    .

J

[johnparker007]

Let me set up and run a longer test over the course of today   

Big thanks again @andrew96for getting some specifics info to go on, hopefully after both versions (original + patched) have run for much longer time, we'll see the variance continue to be present on just the patched ROM.  Then that means I can try to develop a fix without a real machine, which is handy!

I've seen I need to write a enhanced standalone version of the Autoplayer as the one built into MFME doesn't speed up when MFME is accelerated.  So I can't press the buttons fast enough with it!   Last night's £300 tests above were just done with me manually pressing start/0/1 very quickly.

Hopefully I can get enough data built up today...

Edited May 17, 2022 by johnparker007

[serene02]

Very interesting stuff this

J

[vectra666]

I also read somewhere on here via a topic  mfme doesn’t represent the actual machines play 

I mean when you play say cloud999 the mfme player normally puts in £1’s at a time obviously during its play you win £1’s 20ps and tokens 

normal real machine play the player would recycle tokens and change through the machine where as mfme only £1’s

so the machine would hold back token wins as it thinks it’s not been refilled 

 

[serene02]

Where is this Vecs?  There is something about it in the about box.  But those comments were made when around the time of the ‘fairplay’ campaign.  It states that mfme may not represent or emulate accurately.

In general though I believe machines run how they should, a few quirks, a lot of which could be down to a number of factors.  A real machine would be played potentially 1000’s spins a day, when you play a game in mfme you are playing it the same way or just forcing out top features as you have unlimited virtual cash, but they rarely get played normally.

I’m hoping once we get shared ram states and limited funds, machines will get played in various styles.

J 

[vectra666]

48 minutes ago, serene02 said:

Where is this Vecs?  There is something about it in the about box.  But those comments were made when around the time of the ‘fairplay’ campaign.  It states that mfme may not represent or emulate accurately.

In general though I believe machines run how they should, a few quirks, a lot of which could be down to a number of factors.  A real machine would be played potentially 1000’s spins a day, when you play a game in mfme you are playing it the same way or just forcing out top features as you have unlimited virtual cash, but they rarely get played normally.

I’m hoping once we get shared ram states and limited funds, machines will get played in various styles.

J 

Useless info here maybe

 

[johnparker007]

Not much in it so far - the target % is set to 88%.  The one on the right is Chr-free patched:



will leave them running for the rest of the day... they can't play as fast as the other stuff I did recently, but still playing quicker than realtime  

[johnparker007]

Well as more money is going through they are getting closer together in terms of their payout percentage, as one would expect if there is no bug manifesting... so perhaps a real machine is needed to reproduce this issue reported by Andrew's friend - that Cloud 999 Chr-free patched ROMs play badly in terms of payout percentage:


(the one on the right is Chr-free patched ROM, left is original ROM)

Edited May 17, 2022 by johnparker007

[serene02]

Looks like it’s settling down nicely

J

[serene02]

Great work bud, especially if you find that there are no issues with the CHR patched roms you do at least.

Is Andy under the impression that there is something going on when they are being run on actual hardware though?  Still, you understand the assembly code and you don’t see any obvious connections to any skewed percentages….  Yet anyway, be interesting to see the end result.

Ah, I just thought though, there are BWB Cloud 999 versions available I believe!!!  I think Andy’s mate did specify Barcrest Cloud 999 though.

J

[johnparker007]

Right @serene02 @andrew96 I'll call this emulator test complete! 

So, £2,000 into each machine.  Left machine, original ROM; right machine Chr-free patched ROM:

They are very close to each other, though both a little higher than the 88% that apparently they are set to in configuration.

So it seems the Chr-free patch has no effect on payout percentage over a statistically significant sample of games on Barcrest Cloud 999 machines (running CLN 4.0 ROMs).

The only other thing that could be tested, is to perform a 'proper' test on a real Cloud 999 machine.

This would be the same as above:

- fit the ROM
- reset the RAM
- reset the machine's meters
- play £2,000 of credits
- note down the IN / OUT meter values, to calculate the payout percentage over those £2,000 worth of credits

If there is a significant difference between original and chr-free ROM, there is a problem.  If there is no significant difference, then it would seem like there isn't a problem.

Always good to check these things out  

Edited May 17, 2022 by johnparker007

[serene02]

Looks good to me.  These do start off in an extremely good mood from a ram reset that could be why they are both slightly over.  They find there target percentage from zero after a ram reset.  A lot of machines from this era do.

J

[gemini17]

On 27/04/2022 at 17:23, spa said:

I have been trying to contact @Ploggy and @Leeham, with no joy. Does anybody know them at all? Hopefully they are ok!

I used to be in touch with Ploggy regularly on Fruit emu but have not heard from him in a long time.

Tommy C did say to me he had been on dif in the past but I have not sadly seen him.

I too hope he and all past members (who have contributed so much to the continued development of fruit emulation)

, are well.

Jay

[gemini17]

1 hour ago, johnparker007 said:

Well as more money is going through they are getting closer together in terms of their payout percentage, as one would expect if there is no bug manifesting... so perhaps a real machine is needed to reproduce this issue reported by Andrew's friend - that Cloud 999 Chr-free patched ROMs play badly in terms of payout percentage:


(the one on the right is Chr-free patched ROM, left is original ROM)

This is all truly incredible and mesmerising to see, walk around and play(well when I understand how to play a machine-lol)

I am slowly going through the 100's of pages on the topic but sadly with supporting 3 family members with disabilities I'm not quite up to speed.

I'm also not the brightest with any tech John but will get there in the end.

I will send on a donation shortly as I will be sending on my annual donation to this incredible site.

Many thanks to one and all who have contributed to the project and continue to give up many hours of their personal life to keep the site alive in so many ways.

You are all stars with such incredible individual talents.

 

Jay

[johnparker007]

26 minutes ago, serene02 said:

These do start off in an extremely good mood from a ram reset that could be why they are both slightly over.  They find there target percentage from zero after a ram reset.  A lot of machines from this era do.

Ah right, that might explain higher numbers slowly falling over time.  Plus I was only playing this with zero key (one pound coins), no tokens...  still, both ROMs seems to perform exactly the same, so there's certainly no logical issue at least