Forums

http://web.inter.nl.net/users/J.Kortink/sn76489.htm

The exact white noise sequence used in the PSG. It's currently unknown and we use ^=0x18000, but this is the true sequence some genius has found!

Yes, I know, I've a very sad man!

Quote

>
> http://web.inter.nl.net/users/J.Kortink/sn76489.htm

> The exact white noise sequence used in the PSG. It's currently unknown and we use ^=0x18000, but this is the true sequence some genius has found!

> Yes, I know, I've a very sad man!

Update, OG and ASG say it's "1 + x + x^14 + x^16"
What does that mean, the shift pattern is 0x14001 or something?

Quote

> > Yes, I know, I've a very sad man!

Well, I spent an hour or so trying to crack the scheme, by converting it to a bitmap and generating data on the fly with variable settings, but didn't get very far due to having forgotten all the getdibbits() stuff. But I digress.

Quote

> Update, OG and ASG say it's "1 + x + x^14 + x^16"
> What does that mean, the shift pattern is 0x14001 or something?

Who are OG and ASG?

Anyway...

We know it's a 16-bit shift register. And I know that for a 16-bit shift register there's only one feedback value that gets a sequence of 65536 values. The optimal feedback network for a given size of shift register is not mathematically determinable, only by trial and error, and there are tables out there for hardware makers.

So, look here and see the result. I have yet to test it though, and the initial value still needs looking at.

Maxim

Quote

> > > Yes, I know, I've a very sad man!

> Well, I spent an hour or so trying to crack the scheme, by converting it to a bitmap and generating data on the fly with variable settings, but didn't get very far due to having forgotten all the getdibbits() stuff. But I digress.

> > Update, OG and ASG say it's "1 + x + x^14 + x^16"

> > What does that mean, the shift pattern is 0x14001 or something?

> Who are OG and ASG?

Olivier Galibert (working on lots of things in MAME) and Aaron Giles (working on thousands of things in MAME). They are interested in SN76496 for many games in MAME (e.g. System E) and MESS (Many systems, including some obscure thing called the Sega Master System ;). They calculated the algorithm for the PSG noise in just minutes using an algorithm known as Berlekamp-Massey...... these guys are amazing!

The information they have in the Mame source on the PSG is unbelievable by the way - that's how I found out about the shift register for the noise channel. And looks like the sn76496 code just got even MORE accurate!

Quote

> Anyway...

> We know it's a 16-bit shift register. And I know that for a 16-bit shift register there's only one feedback value that gets a sequence of 65536 values.

65535 actually, but I'll let you off just this once ;)

No Text and no TV and no beer makes Homer go crazy

Author	Message
finaldave Guest	Maxim, and perfectionists everywhere, you'll love this... Posted: Wed Mar 20, 2002 9:04 pm
finaldave Guest	http://web.inter.nl.net/users/J.Kortink/sn76489.htm The exact white noise sequence used in the PSG. It's currently unknown and we use ^=0x18000, but this is the true sequence some genius has found! Yes, I know, I've a very sad man!

finaldave Guest	Posted: Wed Mar 20, 2002 9:10 pm
finaldave Guest	Quote > > http://web.inter.nl.net/users/J.Kortink/sn76489.htm > The exact white noise sequence used in the PSG. It's currently unknown and we use ^=0x18000, but this is the true sequence some genius has found! > Yes, I know, I've a very sad man! Update, OG and ASG say it's "1 + x + x^14 + x^16" What does that mean, the shift pattern is 0x14001 or something?

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Hmm? Posted: Thu Mar 21, 2002 6:49 pm
	Quote > > Yes, I know, I've a very sad man! Well, I spent an hour or so trying to crack the scheme, by converting it to a bitmap and generating data on the fly with variable settings, but didn't get very far due to having forgotten all the getdibbits() stuff. But I digress. Quote > Update, OG and ASG say it's "1 + x + x^14 + x^16" > What does that mean, the shift pattern is 0x14001 or something? Who are OG and ASG? Anyway... We know it's a 16-bit shift register. And I know that for a 16-bit shift register there's only one feedback value that gets a sequence of 65536 values. The optimal feedback network for a given size of shift register is not mathematically determinable, only by trial and error, and there are tables out there for hardware makers. So, look here and see the result. I have yet to test it though, and the initial value still needs looking at. Maxim

finaldave Guest	Re: Hmm? Posted: Thu Mar 21, 2002 10:29 pm
finaldave Guest	Quote > > > Yes, I know, I've a very sad man! > Well, I spent an hour or so trying to crack the scheme, by converting it to a bitmap and generating data on the fly with variable settings, but didn't get very far due to having forgotten all the getdibbits() stuff. But I digress. > > Update, OG and ASG say it's "1 + x + x^14 + x^16" > > What does that mean, the shift pattern is 0x14001 or something? > Who are OG and ASG? Olivier Galibert (working on lots of things in MAME) and Aaron Giles (working on thousands of things in MAME). They are interested in SN76496 for many games in MAME (e.g. System E) and MESS (Many systems, including some obscure thing called the Sega Master System ;). They calculated the algorithm for the PSG noise in just minutes using an algorithm known as Berlekamp-Massey...... these guys are amazing! The information they have in the Mame source on the PSG is unbelievable by the way - that's how I found out about the shift register for the noise channel. And looks like the sn76496 code just got even MORE accurate! Quote > Anyway... > We know it's a 16-bit shift register. And I know that for a 16-bit shift register there's only one feedback value that gets a sequence of 65536 values. 65535 actually, but I'll let you off just this once ;)

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14745 Location: London	Continued on S8-dev nt Posted: Fri Mar 22, 2002 9:50 am
	No Text and no TV and no beer makes Homer go crazy

Forums

View topic - Maxim, and perfectionists everywhere, you'll love this...