Best way of detecting the AGA chipset?

Steve · 24 January 2014, 14:15

Hi. I am wondering what is the best way to detect the AGA chipset? In the howtocode page it gives this listing to detect the AGA chipset. Are there other (better) ways?

Code:

GFXB_AA_ALICE   equ 2
gb_ChipRevBits0 equ $ec

; Call with a6 containing GfxBase from opened graphics.library

   btst  #GFXB_AA_ALICE,gb_ChipRevBits0(a6)
   bne.s is_aa

Steve · 24 January 2014, 14:28

Found some more code.

That's a bit unreliable, it's better to check Agnus/Alice version...

Code:

move.w $DFF004,d0
and.w #$6F00,d0                      ; 0110111100000000
cmp.w #$2200,d0                      ; 0010001000000000
beq.b .aga
cmp.w #$2300,d0                      ; 0010001100000000
.aga: seq.b ISAGA

...
...

ISAGA: dc.b 0

Galahad/FLT · 24 January 2014, 14:41

Theres generally a combination of tests people do, because sometimes one check in itself won't be enough.

1). Check for 2meg of chip ram, basic configuration for AGA, but obviously doesn't stop others having it.
2). Check Kickstart version
3). Check Graphics library version
4). Check processor type

Thats what I did back in the day, but obviously, later on, people started using a lot of this stuff on other Amigas, and that buggered that all up.

hitchhikr · 24 January 2014, 15:01

Code:

isAGA:
                 move.w $dff07c,d0
	         moveq #31-1,d2
	         and.w #$ff,d0
check_loop:      move.w $dff07C,d1
                 and.w #$ff,d1
	         cmp.b d0,d1
            	 bne.b not_AGA
            	 dbf d2,check_loop
            	 or.b #$f0,d0
            	 cmp.b #$f8,d0
            	 bne.b not_AGA
                 moveq #1,d0
                 rts
not_AGA:         moveq #0,d0
                 rts

Steve · 24 January 2014, 15:11

I think this is the code I used when I first started playing with assembler a few years ago. Its taken from the HowToCode2 guide on back2roots ftp. I remember it worked fine when I tested it on my A1200 so what is the difference between this one and the one just posted by hitchhikr? Some comments would have been nice as I'm just learning asm again from where I left off years ago.

Code:

This code will check for AGA

        move.w	$dff07c,d0
	cmp.b	#$f8,d0
	bne.s	.notaga

	; Do your funky AGA Stuff in here!

.notaga

I understand that the above code is checking for the Amiga 1200/4000 Lisa chipset by comparing the returned value. If the returned value is $FC (11111100) then the chipset is ECS otherwise if it returns $F8 (11111000) then the chipset is AGA. Is this is the best way of detecting AGA? I don't understand why hitchhkrs code is so complicated. Is it just being more thorough?

Toni Wilen · 24 January 2014, 15:38

VPOSR (DFF004) AGA check is easiest and 100% reliable test. No extra tests needed like DENISEID/LISAID requires (because it does not exist in OCS Denise).

btw, upper byte of LISAID can be used to detect between A1200 and A4000, it is zero on A1200, non-zero on A4000. (But first check AGA using VPOSR)

hitchhikr · 24 January 2014, 16:58

$dff07c doesn't exist on preAGA architectures so it's supposed to return "bus noise", the routine checks it several times to be sure that it's really an AGA machine ($f8) and not just a random value.

Toni Wilen · 24 January 2014, 20:18

$7c (DENISEID/LISAID) was introduced with ECS Denise. But VPOSR is still the best choice, no need for multiple tests

pandy71 · 24 January 2014, 20:55

Quote:

Originally Posted by Toni Wilen

$7c (DENISEID/LISAID) was introduced with ECS Denise. But VPOSR is still the best choice, no need for multiple tests

But seem to be inconclusive...

http://amiga-dev.wikidot.com/hardware:vposr

8372 (Fat-hr) (agnushr),rev 5 = 22 PAL
8374 (Alice) thru rev 2 = 22 PAL

Toni Wilen · 24 January 2014, 21:06

Quote:

Originally Posted by pandy71

But seem to be inconclusive...

http://amiga-dev.wikidot.com/hardware:vposr

8372 (Fat-hr) (agnushr),rev 5 = 22 PAL
8374 (Alice) thru rev 2 = 22 PAL

I am quite sure it is documentation bug and it should say 21 Pal. (22 does not match NTSC model which is 31)

AmigaOS detects AGA by checking if bit 9 of VPOSR is set.

Steve · 25 January 2014, 20:42

Quote:

Originally Posted by Steve

Found some more code.

That's a bit unreliable, it's better to check Agnus/Alice version...

Code:

beq.b .aga
.aga: seq.b ISAGA

ISAGA: dc.b 0

Thanks for all the input. I will have a play with the VPOSR test. I am curious about this bit of code though. I have looked through the 68k instruction set but can't find a seq.b instruction anywhere. Is this a typo?

Galahad/FLT · 25 January 2014, 20:43

Just checked by R-Type 2 AGA fix from back in the day, and I used the same method as Hitchhiker back then.... i'd forgotten!

Toni Wilen · 25 January 2014, 20:50

Quote:

Originally Posted by Steve

Thanks for all the input. I will have a play with the VPOSR test. I am curious about this bit of code though. I have looked through the 68k instruction set but can't find a seq.b instruction anywhere. Is this a typo?

It is most likely called Scc in documentation. cc = any condition code.

Steve · 25 January 2014, 22:08

Quote:

Originally Posted by Toni Wilen

It is most likely called Scc in documentation. cc = any condition code.

Yes you are right. Thanks again.

24 January 2014, 14:28	#2
Steve I Identify as an Ewok Join Date: Jul 2001 Location: North Lincolnshire Age: 45 Posts: 2,356	Found some more code. That's a bit unreliable, it's better to check Agnus/Alice version... Code: move.w $DFF004,d0 and.w #$6F00,d0 ; 0110111100000000 cmp.w #$2200,d0 ; 0010001000000000 beq.b .aga cmp.w #$2300,d0 ; 0010001100000000 .aga: seq.b ISAGA ... ... ISAGA: dc.b 0 Last edited by Steve; 24 January 2014 at 16:09.

24 January 2014, 15:11	#5
Steve I Identify as an Ewok Join Date: Jul 2001 Location: North Lincolnshire Age: 45 Posts: 2,356	I think this is the code I used when I first started playing with assembler a few years ago. Its taken from the HowToCode2 guide on back2roots ftp. I remember it worked fine when I tested it on my A1200 so what is the difference between this one and the one just posted by hitchhikr? Some comments would have been nice as I'm just learning asm again from where I left off years ago. Code: This code will check for AGA move.w $dff07c,d0 cmp.b #$f8,d0 bne.s .notaga ; Do your funky AGA Stuff in here! .notaga I understand that the above code is checking for the Amiga 1200/4000 Lisa chipset by comparing the returned value. If the returned value is $FC (11111100) then the chipset is ECS otherwise if it returns $F8 (11111000) then the chipset is AGA. Is this is the best way of detecting AGA? I don't understand why hitchhkrs code is so complicated. Is it just being more thorough? Attached Thumbnails Last edited by Steve; 24 January 2014 at 15:41.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Detecting ACA6x0 at startup	jimbob	support.Hardware	18	08 January 2013 22:25
Help on detecting Amiga model from within a script	gulliver	support.Other	5	11 October 2011 18:56
EHB sprites with AGA chipset ?	FrenchShark	Coders. General	4	17 September 2009 06:37
GameBase not detecting WinUAE (or any emulators at all)	yoshimitsuhiro	project.MAGE	1	15 March 2009 21:38
Does "Cycle-exact" do nothing with the AGA chipset?	fmcpma	support.WinUAE	10	26 October 2006 13:11

24 January 2014, 14:15	#1
Steve I Identify as an Ewok Join Date: Jul 2001 Location: North Lincolnshire Age: 45 Posts: 2,356	Best way of detecting the AGA chipset? Hi. I am wondering what is the best way to detect the AGA chipset? In the howtocode page it gives this listing to detect the AGA chipset. Are there other (better) ways? Code: GFXB_AA_ALICE equ 2 gb_ChipRevBits0 equ $ec ; Call with a6 containing GfxBase from opened graphics.library btst #GFXB_AA_ALICE,gb_ChipRevBits0(a6) bne.s is_aa

24 January 2014, 14:41	#3
Galahad/FLT Going nowhere Join Date: Oct 2001 Location: United Kingdom Age: 50 Posts: 8,986	Theres generally a combination of tests people do, because sometimes one check in itself won't be enough. 1). Check for 2meg of chip ram, basic configuration for AGA, but obviously doesn't stop others having it. 2). Check Kickstart version 3). Check Graphics library version 4). Check processor type Thats what I did back in the day, but obviously, later on, people started using a lot of this stuff on other Amigas, and that buggered that all up.

24 January 2014, 15:38	#6
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,506	VPOSR (DFF004) AGA check is easiest and 100% reliable test. No extra tests needed like DENISEID/LISAID requires (because it does not exist in OCS Denise). btw, upper byte of LISAID can be used to detect between A1200 and A4000, it is zero on A1200, non-zero on A4000. (But first check AGA using VPOSR)

24 January 2014, 16:58	#7
hitchhikr Registered User Join Date: Jun 2008 Location: somewhere else Posts: 511	$dff07c doesn't exist on preAGA architectures so it's supposed to return "bus noise", the routine checks it several times to be sure that it's really an AGA machine ($f8) and not just a random value.

24 January 2014, 20:18	#8
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,506	$7c (DENISEID/LISAID) was introduced with ECS Denise. But VPOSR is still the best choice, no need for multiple tests

25 January 2014, 20:43	#12
Galahad/FLT Going nowhere Join Date: Oct 2001 Location: United Kingdom Age: 50 Posts: 8,986	Just checked by R-Type 2 AGA fix from back in the day, and I used the same method as Hitchhiker back then.... i'd forgotten!

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