Hidden Amiga 68000 opcode?

pmc · 19 April 2007, 20:06

I've been watching the Razor 1911 demo Vertical Insanity.

The scrolltext talks about a hidden 68000 instruction that can be used on the Amiga only to move words into the custom chip registers called movef - 68000 opcode $3222. It claims that this move command does this very quickly.

Does anyone know if this is true? If it is, is it actually any use, and how would I go about using it from my assembler? Devpac 3 complains about an unrecognised instruction if I type in movef...

musashi5150 · 19 April 2007, 20:15

That's a hoax my friend

That opcode is pure fiction. Their bars are done using a HAM trick.

pmc · 19 April 2007, 20:26

Those cheeky scamps! I had a feeling it would turn out to be bullshit... At least I know - nice one musashi5150 - hey, hang on a minute, I spelt your name right!

Toni Wilen · 19 April 2007, 21:18

Easiest way to "confirm" this is by trying to disassemble $3222...

alexh · 20 April 2007, 09:11

Quote:

Originally Posted by pmc

I've been watching the Razor 1911 demo Vertical Insanity.The scrolltext talks about a hidden 68000 instruction that can be used on the Amiga only to move words into the custom chip registers called movef - 68000 opcode $3222. It claims that this move command does this very quickly.

There is a C64 demo homage (or was it the other way around?) In the scroll text they too have found a "secret opcode" on the C64

http://www.pouet.net/prod.php?which=15401

You need to read the scroll text in the secret part of the Razor 1911 Amiga Vertial Insanity demo. Hold right mouse button while the demo decrunches for an alternative scroller and a rastertime counter.

pmc · 20 April 2007, 16:16

Hmmm, interesting... Thanks for the info alexh.

Quote:

You need to read the scroll text in the secret part of the demo. Hold right mouse button while the demo decrunches for an alternative scroller and a rastertime counter.

I'll try that out.

pmc · 20 April 2007, 19:43

Just accessed the hidden part and read the real explanation for how the 'vertical copper bars' were done, very interesting and quite clever really...

mr_0rga5m · 22 April 2007, 15:50

Right mouse on the c64

.

pmc · 22 April 2007, 17:44

@mr_0rga5m - err... no. When I first read alexh's post I read it like that. He means press the rmb when the Amiga demo's loading...

oRBIT · 08 May 2007, 18:34

The 6502 processor actually had quite a few more or less undocumented opcodes which did lots of wierd stuff. Wonder if this could be the case on the 680x0 aswell..? hmm..

alexh · 08 May 2007, 20:37

No. They were taking the piss. Joking, fooling around.

pmc · 08 May 2007, 21:58

In the demo they were joking but could oRBIT have a point - aside from the bullshit in the demo are there actually any undocumented opcodes?

musashi5150 · 09 May 2007, 09:02

I don't think there are... but Toni Wilen would be the man to ask

There was a thread along a similar line about quirky 68K aspects a while back, http://eab.abime.net/showthread.php?t=23315 might be worth a little look if you're interested.

pmc · 09 May 2007, 10:54

Thanks mate, as usual, you are 'da man'!

Toni Wilen · 09 May 2007, 15:34

I am quite sure the answer is "no".

68000+ (and all "modern" CPUs) have illegal instruction exception (or similar feature), 6502 don't. Another difference is that 68000 internally executes microcode (every 68k instruction is internally executed using few internal basic microcode instructions)

Check 68000 patents, they are most interesting, they even include (very unreadable) microcode listing.(http://patft.uspto.gov/netahtml/PTO/srchnum.htm Patent numbers: 4342078 and 4296469)

alexh · 09 May 2007, 16:21

Ooooo.

I once did a 68000 in VHDL but I did it as a non microcoded state machine.

It was a lot bigger (perhaps 8x or 9x the size) than the 68000 transistors (17k gates).

To match timing on some instructions I had to do some horrible cycle wait states.

It was never even close to cycle accurate, much like the existing attempts at a 68000 in HDL.

StrategyGamer · 10 May 2007, 07:13

@Alexh

I don't care how many gates it uses. I just wanna know how fast is it?

Can it go faster than 50 Mhz?

I don't want it to be cycle accurate to a slow lame 68000 from 1978.

alexh · 10 May 2007, 11:45

Quote:

Originally Posted by StrategyGamer

@Alexh
I don't care how many gates it uses. I just wanna know how fast is it?
Can it go faster than 50 Mhz?

According to Synopsys it could run at over 300MHz at 0.13um. All I need now is $2,000,000 for a set of masks

Quote:

I don't want it to be cycle accurate to a slow lame 68000 from 1978.

I do... otherwise Amiga stuff (and especially Atari ST stuff) wont work.

StingRay · 13 May 2007, 22:52

Quote:

Originally Posted by Toni Wilen

Easiest way to "confirm" this is by trying to disassemble $3222...

What is this supposed to confirm? They spoke about an opcode they have found, not some memory location. It's the same as if you would try to disassemble $4e71 and expect to find nop instruction there...

pmc · 14 May 2007, 08:43

@StingRay - I think Toni means that $3222 is the hex value of the machine language opcode rather than a memory location... For example, a real instruction like:

moveq #1,d1

is actually encoded as

0111001000000001 (or $7201 in hex)

by your assembler for the processor to then execute...

At least I think that's what Toni means!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
hidden Amiga gems/games	s2325	Retrogaming General Discussion	209	12 October 2017 12:11
Found hidden amiga mouse pointer in James Brown Is Dead music video!	Hewitson	Nostalgia & memories	6	18 May 2010 11:21
New Flash Game (with amiga stuff in it, some hidden)	Ironclaw	Retrogaming General Discussion	6	26 April 2009 21:18
New opcode for 68000 family	clenched	request.UAE Wishlist	15	14 April 2009 15:02
Amiga, 68000 and CPU speed	SilentBob	Retrogaming General Discussion	3	15 October 2006 15:45

19 April 2007, 20:06	#1
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	Hidden Amiga 68000 opcode? I've been watching the Razor 1911 demo Vertical Insanity. The scrolltext talks about a hidden 68000 instruction that can be used on the Amiga only to move words into the custom chip registers called movef - 68000 opcode $3222. It claims that this move command does this very quickly. Does anyone know if this is true? If it is, is it actually any use, and how would I go about using it from my assembler? Devpac 3 complains about an unrecognised instruction if I type in movef...

19 April 2007, 20:15	#2
musashi5150 move.w #$4489,$dff07e Join Date: Sep 2005 Location: Norfolk, UK Age: 42 Posts: 2,351	That's a hoax my friend That opcode is pure fiction. Their bars are done using a HAM trick.

19 April 2007, 20:26	#3
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	Those cheeky scamps! I had a feeling it would turn out to be bullshit... At least I know - nice one musashi5150 - hey, hang on a minute, I spelt your name right!

19 April 2007, 21:18	#4
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,529	Easiest way to "confirm" this is by trying to disassemble $3222...

20 April 2007, 19:43	#7
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	Just accessed the hidden part and read the real explanation for how the 'vertical copper bars' were done, very interesting and quite clever really...

22 April 2007, 15:50	#8
mr_0rga5m Tik Gora :D Join Date: Oct 2001 Location: Round yo momma's Posts: 1,273	Right mouse on the c64 .

22 April 2007, 17:44	#9
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	@mr_0rga5m - err... no. When I first read alexh's post I read it like that. He means press the rmb when the Amiga demo's loading...

08 May 2007, 18:34	#10
oRBIT Zone Friend Join Date: Apr 2006 Location: Gothenburg/Sweden Age: 48 Posts: 344	The 6502 processor actually had quite a few more or less undocumented opcodes which did lots of wierd stuff. Wonder if this could be the case on the 680x0 aswell..? hmm..

08 May 2007, 20:37	#11
alexh Thalion Webshrine Join Date: Jan 2004 Location: Oxford Posts: 14,367	No. They were taking the piss. Joking, fooling around.

08 May 2007, 21:58	#12
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	In the demo they were joking but could oRBIT have a point - aside from the bullshit in the demo are there actually any undocumented opcodes?

09 May 2007, 09:02	#13
musashi5150 move.w #$4489,$dff07e Join Date: Sep 2005 Location: Norfolk, UK Age: 42 Posts: 2,351	I don't think there are... but Toni Wilen would be the man to ask There was a thread along a similar line about quirky 68K aspects a while back, http://eab.abime.net/showthread.php?t=23315 might be worth a little look if you're interested.

09 May 2007, 10:54	#14
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	Thanks mate, as usual, you are 'da man'!

09 May 2007, 15:34	#15
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,529	I am quite sure the answer is "no". 68000+ (and all "modern" CPUs) have illegal instruction exception (or similar feature), 6502 don't. Another difference is that 68000 internally executes microcode (every 68k instruction is internally executed using few internal basic microcode instructions) Check 68000 patents, they are most interesting, they even include (very unreadable) microcode listing.(http://patft.uspto.gov/netahtml/PTO/srchnum.htm Patent numbers: 4342078 and 4296469)

09 May 2007, 16:21	#16
alexh Thalion Webshrine Join Date: Jan 2004 Location: Oxford Posts: 14,367	Ooooo. I once did a 68000 in VHDL but I did it as a non microcoded state machine. It was a lot bigger (perhaps 8x or 9x the size) than the 68000 transistors (17k gates). To match timing on some instructions I had to do some horrible cycle wait states. It was never even close to cycle accurate, much like the existing attempts at a 68000 in HDL.

10 May 2007, 07:13	#17
StrategyGamer Total Chaos AGA is fun! Join Date: Jun 2005 Location: USA Posts: 873	@Alexh I don't care how many gates it uses. I just wanna know how fast is it? Can it go faster than 50 Mhz? I don't want it to be cycle accurate to a slow lame 68000 from 1978.

14 May 2007, 08:43	#20
pmc gone Join Date: Apr 2007 Location: completely gone Posts: 1,596	@StingRay - I think Toni means that $3222 is the hex value of the machine language opcode rather than a memory location... For example, a real instruction like: moveq #1,d1 is actually encoded as 0111001000000001 (or $7201 in hex) by your assembler for the processor to then execute... At least I think that's what Toni means!

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)