What Exactly Does Clearing The RAM DO, And Does It Affect The RNG?

[brichter]

I guess they still use S+es in PA...

[StatFreak]

Stat, can you explain the following statement?

Regarding the assertion made by many techs that these machines are more likely to hit a top prize after a clear, or within some specified time frame following a clear (I have heard stories ranging from 24 hours to 168 hours), there is a possibility  that the RNG might  be more likely to choose certain number combinations early in its cycle following a reseed.

Does a RAM clear change the seed of the RNG?

That question is best answered by Stolistic.  

I was speaking to the urban legend (or perhaps, as yet unproven truth ) that these machines cough up more jackpots immediately following a clear than at other times.

The thing that sticks in my craw about this debate is that even IF (and it's a big IF) the RNG somehow proves to be less random after an initial seed, it's not clear to me how that translates to jackpots when different games would require different RNG outcomes to yield their top prizes.

As I understand it, the RNG uses bit shifting against a polynomial, similar to the way that checksums are calculated, to produce double precision floating point values of n where 0 >= n < 1. In actuality, these numbers are represented in binary, of course.

As a programmer, if I want an integer between 0 and 63 for a 64 stop machine, there are a few ways to get a result. One way would be to multiply the random number returned by 64 and truncate the result. A person would not want to multiply by 63 and round the result, because the values of 0 and 64 would be half as likely to occur that way. Another (and better IMO) way would be to multiply the random number returned by an arbitrarily large number, perform Modulo division by 64, and take the remainder.

As has been stated earlier in the thread, the machine runs through random values between 0 and 1 tens of thousands of times per second or more as long as the machine is powered and in idle mode. On an S-Plus, the computer polls the RNG for three values at the instant that the player takes the first action to initiate a game (presses "bet max", presses "bet one", or drops a coin into the comparitor that is accepted).  The player’s action is the truly randomizing event.

The computer program looks at the reel chip in a specific memory location for a single hex value that represents the number of virtual stops for that game, which the program uses to convert the three random numbers into "integers" between 0 and n, where n is the number (in hex) found in the reel chip, by using one of the standard approved methods such as those that I mentioned above (most likely the modulo division, but perhaps something else. It doesn't matter much as long as the method creates evenly distributed integers across the range of desired results.) Let's call the three integers I, J, and K.

The program then looks up position I in the reel chip virtual stop table listing for strip number one, J in the virtual stop table for strip 2, and K in the virtual stop table for strip 3. At this point, the program STILL doesn't know what the "symbols" are or what the outcome is! The values retrieved from the virtual stop tables are pointers that reference another table containing the numbers that correspond to the physical stop “symbols�.

I'm not going to take this further, but suffices to say that in order to hit the jackpot, the RNG has to return three values in order which, when converted into integers covering the range of virtual stops for the chip installed, yields the three integers matching the virtual stops that correspond to the jackpot symbols.

Consider this scenario:
1. A game with 64 virtual stops, like RWB. SS4633 with one winning combination.
    The magic numbers (in decimal) are 24, 24, 23.

2. The same game with 200 virtual stops. RWB SS7115 with two winning combinations.
    The magic numbers (in decimal) are 62, 65, (67/68)

3. Another game with 64 stops, Haywire SS4079 with eight winning combinations.
    The magic numbers (in decimal) are (10/46), (9/43), (11/45)

In each case, the RNG returned 0.574168647, 0.900696219, 0.037847008


I’m going to arbitrarily multiply each random number by 1,000,000, take the truncated integer and do modulo division.

For the machines with 64 stops, I get:
574168 / 64 = 8971 remainder 24.
900696 / 64 = 14073 remainder 24.
37847 / 64 = 591 remainder 23.

Well what do you know? I got the jackpot on the RWB SS4633. Unfortunately, I got nothing on the Haywire SS4079.

For the machine with 200 stops, I get:
574168 / 200 = 2870 remainder 168.
900696 / 200 = 4503 remainder 96.
37847 / 200 = 189 remainder 47.

Nothing here for the 200 stop RWB7115. Sorry folks.

The same RNG outcome produced three markedly different results for three different chips.

Now explain to me again how the RNG must have some limited pattern during it’s first hours after a clear that leads to more jackpots..

[stayouttadabunker]

I dunno...it just does!...LOL

[stayouttadabunker]

just kidding ya Stat...

Thanks for giving us your mathematical viewpoint on the way the RNG works.
That post and this entire thread should be a sticky because something else
will be discovered within our lifetimes that will shed more light on this I'm sure...

It would be neat for someone like stolistic or channelmaniac
to give us their technical viewpoint on how the RNG works...
combine the two and we will have the answer!

[edski]

This first video was on track but the second video...LOL She's a bloody liar!!!!

WHAT!!!
Please explain!!!!
And while you're at it, please explain why Al Gore is your hero?

[stayouttadabunker]

Who's Al Gore? LOL

She's probably alright...she's just a pretty face reading a script given to her...

[brichter]

She states they must change a chip to change the percentage.

I'm attaching the base program paytable for an SB000070, note it shows 10 different paytable IDs with 10 different percentages from 83.76% all the way to 98.05%. All of these are selectable via the menu, no chip change necessary. I think that's what Bunker was referring to when he stated she was lying.

I doubt seriously she was lying, she just didn't have a clue as to the veracity of the facts she was given.

<edit> forgot the attachment...

[reho33]

I do not know when these videos were made. The PGCB is like the DMV, control 2010 with 1985 rules!

[brichter]

The PGCB is like the DMV, control 2010 with 1985 rules!

 

[stayouttadabunker]

Thank you for clearing that up Brichter...!
You were right...she was not lying...she just doesn't know.
I was just exaggerating the video and her points...
like I said - "just a pretty face reading a script"...
Reho's got a point there though...
The video may have been made before the multi-paytable chips came out.
or, she just plain didn't know about them.
Can anyone tell from the video what year those machines were she was pulling the handle on?

[brichter]

Looks like newer technology than S+, at any rate...

[stayouttadabunker]

Yea! It looks like a "Monopoly" standalone machine behind her...
Look at that yellow bill acceptor...fairly modern...is that machine made by Williams?

[Thor777]

Thor777, stayouttadabunker, edski, brichter, Buzz, Neonkiss, lonnie, uniman, ROCKET, lowflight, TZtech, theDotster, slots4home, Stolistic, harp, Ozzy, Bettor Slots and 12 Guests are viewing this topic.

This is a HOT topic !  ...  It doesn't get much better than this! ... A thread to be remembered 

  You guys are WAY beyond me in this discussion... I am going to drop out and watch the Fallout 

[StatFreak]

Thor777, stayouttadabunker, edski, brichter, Buzz, Neonkiss, lonnie, uniman, ROCKET, lowflight, TZtech, theDotster, slots4home, Stolistic, harp, Ozzy, Bettor Slots and 12 Guests are viewing this topic.

This is a HOT topic !   ...  It doesn't get much better than this! ... A thread to be remembered  

  You guys are WAY beyond me in this discussion... I am going to drop out and watch the Fallout  

Now it's only:

StatFreak, edski, brichter, Buzz, Neonkiss, lonnie, uniman, ROCKET, lowflight, TZtech, theDotster, slots4home, Stolistic, harp, Ozzy, Bettor Slots and 9 Guests are viewing this topic.

The topic seems to be fizzling out...    



<EDIT>
Oh wait, Thor777 and stayouttadabunker came back. That makes it:

StatFreak, stayouttadabunker, Thor777, edski, brichter, Buzz, Neonkiss, lonnie, uniman, ROCKET, lowflight, TZtech, theDotster, slots4home, Stolistic, harp, Ozzy, Bettor Slots and 9 Guests are viewing this topic.

Maybe my fizzle drizzle was premature. Now if we could just get those guests back... 

[stayouttadabunker]

I know...we ran outta viewpoints...
The east coast is going to bed...
It was fun really! g'night!

[StatFreak]

I know...we ran outta viewpoints...
The east coast is going to bed...
It was fun really! g'night!

THAT's IT!! The RNGs on the East coast are more conservative than the liberal RNGs on the West coast!! I knew  that there had to be an explanation.
 

Oh, wait. It's the Easterners that are saying that the RNG is more liberal when it's young, and the Westerners that are saying it's not.    
That shoots that theory all to Hell.

[jay]

PA only got slot parlors about 2 years ago so the multi-denominations and variable percentages were certainly available.

Before that if you wanted to play slots you had to drive to Wheeling West Virginia. From the north side of Pittsburgh this was about a 25min drive.

The truth behind the scenes is that there is no magic buton in anyones office and the effort to manipulate a entire slot parlors %% payback for the weekend would kill off all profit due to the labor involved.

This does however poise the question: I wonder if SAS could be used to change slot settings without a floor visit ? I know it can be used for coupon validation (for TITO in) so the communication is bi-directional. I really want a casino management platform for my basement !!

[stayouttadabunker]

... I really want a casino management platform for my basement !!

Wait!...Lemme look in my files...Nope, I don't have that one...

My guess is that program doesn't come cheap!  


I want one too!

[Stolistic]

OK here's the deal:

Everything Stat has said, is correct (for the S+ and PE+).  The processor on the S+ is running a RNG via the same software found on the program chip.  The 8032/52 processor does not have any sort of native way to get a random number via a special opcode or seed.  This means the random numbers produced are strictly from a mathematical formula found on the program chip.

I have written my own RNG in the PE+ (which is the same processor as the S+).  The only way to do so, is via bit shifting and use of the parity bit to allow for some nice wild fluctuations.

There is no way to seed (or produce a starting number) based off of the system time or such, since there is no such thing on this processor.  That means all random numbers begin from the same starting point (on a ram clear).  Additional starting points when the machine is restarted etc, come from the last saved number in RAM.

If you could record the first 50 numbers from the RNG, you would see they are always the same "predictable" numbers based on a formula.  Like Stat said, these are derived from bit shifting a number over and over again.  Now realize (like Stat said), these numbers are changing thousands of times per second.  So granted, two machines with identical software, and clock crystal speeds, fully ram cleared would produce the same numbers (if recorded).  But a human being could never press the button at the exact same time to produce the same results.  And getting two machines to run at the same speed is also nearly impossible.  Writing results to RAM during a spin etc could be slightly slower on one or the other and there are just too many components on the board that allow for variances.  And the RNG is always running, via processor interrupt routines.  Pressing the button does not make a routine run to pick the next number, it just looks at a specific location in memory which has been changing thousands of times per second and fetches that value.

So to answer 1 question.  Does a RAM clear affect the RNG?  The answer is Yes.  In the grand scheme of things, does it matter? No.  You will never produce the same spins back to back, no matter how long you try.

[StatFreak]

Thank you Stolistic for filling in the final piece.

[brichter]

OK here's the deal:

Everything Stat has said, is correct (for the S+ and PE+).  The processor on the S+ is running a RNG via the same software found on the program chip.  The 8032/52 processor does not have any sort of native way to get a random number via a special opcode or seed.  This means the random numbers produced are strictly from a mathematical formula found on the program chip.

I have written my own RNG in the PE+ (which is the same processor as the S+).  The only way to do so, is via bit shifting and use of the parity bit to allow for some nice wild fluctuations.

There is no way to seed (or produce a starting number) based off of the system time or such, since there is no such thing on this processor.  That means all random numbers begin from the same starting point (on a ram clear).  Additional starting points when the machine is restarted etc, come from the last saved number in RAM.

If you could record the first 50 numbers from the RNG, you would see they are always the same "predictable" numbers based on a formula.  Like Stat said, these are derived from bit shifting a number over and over again.  Now realize (like Stat said), these numbers are changing thousands of times per second.  So granted, two machines with identical software, and clock crystal speeds, fully ram cleared would produce the same numbers (if recorded).  But a human being could never press the button at the exact same time to produce the same results.  And getting two machines to run at the same speed is also nearly impossible.  Writing results to RAM during a spin etc could be slightly slower on one or the other and there are just too many components on the board that allow for variances.  And the RNG is always running, via processor interrupt routines.  Pressing the button does not make a routine run to pick the next number, it just looks at a specific location in memory which has been changing thousands of times per second and fetches that value.

So to answer 1 question.  Does a RAM clear affect the RNG?  The answer is Yes.  In the grand scheme of things, does it matter? No.  You will never produce the same spins back to back, no matter how long you try.

Outstanding explanation!    

Now, to the last questions.

1) Does the content of RAM/CMOS changing have any effect onn the way the machine performs, and could a RAM clear make the machine behave differently as far as performance (or any of the other adjectives used earlier)? I don't see how this would be possible, since those addresses are well-defined, unlike a PC's  browser cache which can grow or shrink as needed. it would seem that the RAM and CMOS addresses are fixed, and the only thing that changes are the values stored there.

2) Is it any more likely to hit a jackpot just after a RAM clear? From the discussions we've had in the past, I don't see how this would be possible, but there are some who have experiences that would seem to contradict this.