Amiga questions you've always been too embarrassed to ask

[roondar]

Originally Posted by xanderbeanz

1. Please give me a brief desecription of exactly what ARexx is...
2. Why can the Amiga only address 2MB Chip RAM, not enough traces on the motherboard?
3. How is it possible to format an Amiga CF card on WinUAE through a USB CF hub? That blows my mind...I know hard drive formatting is kind of OS controlled, but why does it work? The Emulator is talking through a USB hub on Windows...I just can’t compute...
4. From a theoretical point of view, could a aoundcard be built with 4 or 8 Paula’s on it? Would anyone besides me want this crazy thing?

Thanks!

1. A scripting language that allows cross-process (=cross application) communication. For example, an AREXX script could ask your e-mail client if a specific mail has arrived and if so, tell a different program to do something with the e-mail.
2. The custom chip controlling Chip RAM wasn't made with the ability to address more than 2MB worth. This is not exactly the same as not enough traces, but it's kind of similar in a way.
3. Formatting a CF card can be done because the CF card (and USB hub) don't actually care what data is read or written on the card. In essence, formatting mostly means putting the right data in the right spot so that the OS (in this case Amiga OS) can recognise it as a valid disk to use for writing files (it's slightly more complicated, but this is close enough I think).

Also, from the perspective of Windows there really is no difference between WinUAE and say Word, or MS Paint when it accesses a USB drive. As long as Windows lets applications access USB on the level needed for formatting a CF card, WinUAE can do its thing.
4. Sure, though it might be complicated due to how Paula expects data to be fed into it. But it should be possible.

Then again, the WHY!? question is quite apt here

[Gorf]

1. (A)REXX is a programming language. It is "Turing-complete" so you can (in theory) program everything in it. It's stronghold is remote-controlling other programs - these programs need a so called "Arexx-Port" to make this possible.

2. there are some A4000 boards with more address-lines, to enable like 8MB of ChipRAM ... but there never was a Alice (or Agnus) that would support this. You can configure more ChipRAM in emulation.

3. the device is forwarded as "raw" - a block-device. The OS (in this case AmigaOS) takes care of organizing the data (filesystem).

4. might be very difficult for Paula because it depends on Agnus/Alice - but there are boards and devices with multiple SID-chips

[Paulee_Bow]

Ooh I got two for the price of one there

[ajk]

As for 4 - it would be a cool challenge for sure but the Paula does not really offer any particularly unique features. The SID does and there are many projects around it

[Paulee_Bow]

Originally Posted by ajk

As for 4 - it would be a cool challenge for sure but the Paula does not really offer any particularly unique features. The SID does and there are many projects around it

Paula does AM/FM between channels. That’s pretty unique, especially as you can do it with samples. It’s also 8bit and grungy. The synth community love grungy 8bit samplers.

[NorthWay]

Originally Posted by xanderbeanz

2. Why can the Amiga only address 2MB Chip RAM, not enough traces on the motherboard?

As has already been said, no chips have been made with pins to control more than 2M.


Next up is the layout of the Amiga memory map: The original design only has a 16M address space. The first 512K was marked for chipram and then nothing before Zorro II starts at the 2M mark. The Zorro II space takes up 8M before you get into "Ranger"(slow - well not really, AFAIK the first A1000 expansions using this was actually proper Fastmem) memory space that is nearly another 2M in size. Assorted chips and ROM space then fills the rest.
So where do you put chipmem past 2M? You can eat out of the Zorro II space - that would max you out at near 12M chipmem.
However, the memory lists in an Amiga should support more than one region marked as chipmem so it should be possible to simply add another memory area (somewhere in 32-bit space) and the system would simply work except with more chipmem. This would need a way to detect and add this memory (a custom Kickstart preferably), and from a hw standpoint I would guess that you would design it as a single chipmem range but with the first 2M showing up twice in the address space.
You can already try having more chipmem with Minimig units and WinUAE AFAIK.

[Paulee_Bow]

Thanks

I have another question...

1. Vampire II runs an FPGA that models various custom chips (including super AGA theoretical graphics) and a theoretical 100mhz 68080. Once these design choices became somewhat stable, If someone were crazy enough with millions of ££££ to spend on crackpot projects, could they make these chips for real? Leading to question number 2.

2. With said theoretical ££££ to waste, like millions, could someone make a 68k Amiga that could compete with modern PC’s, maybe gigahertz in speed, that could was 100% compatible with code from the original 68k line, not PPC based. Are there limitations to the 68k architecture in terms of heat and power consumption that couldn’t be cracked? Is this all pointless discussion now FPGA’s exist?

I like exploring the hypotheticals

[Gorf]

1. Yes

2. Yes

At least according to Gunnar his 080 should be able to reach around 1GHz as a ASIC without major redesign standing in the way ... going for multiple cores like AMD should also be possible - there is nothing special to 68k in terms of power consumption: it would face the same difficulties as all other architectures, that can be solved by the same solutions and tricks... given enough money and resources.

It is not pointless since a ASIC will always be faster than a FPGA (for the foreseeable future) - so the idea is kind of intriguing.
we will see if maybe a combined Amiga, Atari, Classic Mac Kickstarter could provide enough money for such a project one day?
(I am expecting costs for ASICs to go way down next decade ...)

[idrougge]

Keep in mind that the original PC processor (808&/8088) is vastly inferior to the original 68000, yet it has managed to evolve into the i5/i7 of today. You could apply the same evolution to the 68000, but not without submitting to the same compromises as the PC has gone through, thereby limiting compatibility with the original CPU — especially at the hardware level.

[Marle]

Originally Posted by xanderbeanz

2. With said theoretical ££££ to waste, like millions, could someone make a 68k Amiga that could compete with modern PC’s, maybe gigahertz in speed, that could was 100% compatible with code from the original 68k line, not PPC based. Are there limitations to the 68k architecture in terms of heat and power consumption that couldn’t be cracked? Is this all pointless discussion now FPGA’s exist?

I like exploring the hypotheticals

The other answers are great but on this point, when the 68060 became available around mid-1995 for the Amiga with the original Phase 5 Cyberstorm/060 it's worth noting that the '060 was faster than Pentium's of equivalent clock speed and then some, so long as it was doing integer workload, unfortunately whilst my memory is somewhat vague, floating point is where the '060 fell behind the Pentium. But in general use a '060 50MHz was when not rendering or doing anything that required fractional calculations was comparable and well place against a 66MHz or maybe even 75MHz Pentium. Of course don't forget Motorola was part of the AIM alliance (Apple IBM Motorola) pushing for PowerPC, so from Motorola's perspective the 68k was no longer a high performance workstation CPU of the future.

The 68000 in 1979 did steal a march on Intel, but they caught up quick, the 286 wasn't really much better than a 68000 but crucially Intel had the 80287 (the FPU) by 1982, whereas Motorola took until 1984 to have the 68020 and 68881 (the first 68k FPU) which was a gap of reasonable proportion, and even then there were issues with limited supply. By 1985 Intel had the 386... and from then on...

What propelled the 68k's initial dominance was that it was just so much better than using an Intel 8088 on paper and in reality but Intel caught up quick and had surpassed the 68k's performance pretty much by the early 90s, arguably earlier when most applications of the 68k series in desktop computing were all 68000 until the Mac II in 1987 with the 68020 and the A2500 in 1988/89.

Essentially this long waffly answer is - there is no reason at all why the 68250 as it may now be could not run as fast as screaming i7, except that its market acceptance in desktop computing is what did for it in terms of securing Motorola's investment in pushing it further and further. Even by the early 90s it was clear the 68000 series was coming to an end, arcade boards were relegating the 68000 to being a network controller or similar, audio was a common application, and Apple were going PPC, so back in the day the future was PA-RISC, Intel x86, MIPS, DEC Alpha, PPC and maybe to some extent Hitachi SH series CPUs. And we all know how that panned out.

Maybe a slightly more philosophical answer. Just the same as why the Amiga custom chips can't access more than 2MB, the original memory map of OCS if you look at it states 512KB for custom chip access with 1.5MB reserved (which became accessible with later Agnus chips), but the original 24bit address space only allowed up to 16MB of addressable memory, but once Amigas could address 32bit memory space with a full fat 68020 or 030 or better, there was no reason other than in the case of Alice in the AGA chipset weren't upgraded to do so.

My speculation has been that Commodore may have had some intention of allowing Alice with a later revision to access the Zorro II memory space for extended chip RAM up to 8MB, but this is just pure conjecture on my part and the very man (Dave Haynie) who created these machines has never said anything of the sort, but looking at the original Amiga memory map vs AGA and Commodore's commitment to 32bit Amigas by mid 1993, this seemed a logical way forwards as Zorro II memory space would become redundant with Zorro III cards.

I think

Vicky

p.s. and regardless of 68k's demise, even Commodore was moving away from it by 1993, with PA-RISC stated as their future CPU architecture. So the 68k's writing was on the wall way back then.

[NorthWay]

Originally Posted by xanderbeanz

If someone were crazy enough with millions of ££££ to spend on crackpot projects, could they make these chips for real?

I'm pretty sure the RISC-V stuff that Hennessy and Patterson talked about during their Turing Award ceremony said they had done ASICs for $14,000 for lots of 100. So $140 per ASIC (I don't know what kind of packaging that is, but you'd guess it would be regular chips for some kind of socket type).

That is well in range of Kickstarter projects...

[NorthWay]

Originally Posted by Marle

there is no reason at all why the 68250 as it may now be could not run as fast as screaming i7

They both have the same problem that the native instructions have to be broken down into smaller steps (the x86 even worse so I believe - though not as many levels of indirection as the PDP-11, it still maxes out at 15 bytes long opcodes AFAIK), but this is something Intel has shown that if you just throw enough manpower and $$$ at it then it will stick in the end.
You should be able to apply the exact same methods to the 68000 line and get just as good results (new streaming opcodes excluded). Which might be what Gunnar has already done - after some time what was more like black magic becomes achievable for mere mortals.

[Paulee_Bow]

Thanks for answering my hypothetical question everyone

[nogginthenog]

Originally Posted by Marle

My speculation has been that Commodore may have had some intention of allowing Alice with a later revision to access the Zorro II memory space for extended chip RAM up to 8MB, but this is just pure conjecture on my part and the very man (Dave Haynie) who created these machines has never said anything of the sort, but looking at the original Amiga memory map vs AGA and Commodore's commitment to 32bit Amigas by mid 1993, this seemed a logical way forwards as Zorro II memory space would become redundant with Zorro III cards.

The A4000 has non-working a 8MB chip ram jumper so it was definitely planned at some point.

[Daedalus]

Yeah, the simplest way of increasing chip RAM would be to use the Zorro-II space, which is likely how the planned A4000 solution would have worked, and how WinUAE works with its 8MB solution. If there were enough address lines available, it would be doable. Zorro-II space wouldn't be available then for Zorro-II peripherals, but I guess the hope would have been to drop Zorro-II compatibility on future machines and go with Zorro-III only.

Others have answered the ARexx question, I'll just add to it by saying that if you imagine VBScript from Microsoft office, but apply it across the entire OS and almost every application seamlessly, that's what ARexx is. Even Workbench itself supports it from OS 3.5 on, allowing a program to add menu items or control windows or icons, even if it was never originally written with that capability. Another example, I had a TV card setup on my A1200 Tower, a remote control receiver and a small LCD screen. Using ARexx, I was able to get the remote receiver to control the TV viewing program, change channels, adjust volume, and give feedback on that on the LCD screen (e.g. "Channel 1: BBC 1", "Volume: 20" etc.) It also cycled the display around, showing the time and date, how many unread emails I had in YAM, and the currently playing song title, be it MP3 or CD. All done using ARexx to connect everything together.

[Daedalus]

Originally Posted by turrican3

But i have just one question with all that hard work and really hard work what is it so hard to cut music in a game and replace it with cdda ??
If it's for the love of paula i understand but some games doesn't have even music, it could be great to make cd32 cdda versions ?? no ?
Why it seems so hard ?? When, so many things, and some, we thought nearly impossible were done... Perhaps whdload could make it in the future in a easy way to patch the games.
It's not a critic, it's really a question ??
Is it because : the way to launch cdda is badly documented ?? what's the stone in front of the path ?

CDDA is a little more complicated, and involves a totally different approach of sending and receiving SCSI commands from the device driver that handles the CD-ROM. This is easy enough to do with system-friendly applications and games, but when the system is shut down, it's made a lot more complicated, potentially even needing different code depending on the interface used.

CD32 versions on WHDLoad are typically hardcoded for the CD32's device driver, which is different to other device drivers found on other Amigas. Also, the nature of WHDLoad is to run the games without requiring the original media, so typically the CD isn't present on the system (and often the system doesn't even have a CD drive). Streaming the audio from the hard drive is even more complicated again when you're dealing with non-system-friendly applications, and the amount of uncompressed data involved would mean a significant proportion of the Amiga's bus time would be used up streaming the audio, meaning big slowdowns in a lot of games with that approach using the standard built-in interfaces.

[Marle]

Back in the day I created a headerless (no interface) Email client for the Amiga that had an ARexx interface meaning any application could send emails (with Base64 encoded attachments), and receive emails and read them and any text attachments e.g. process information. I never got to the stage of say detaching an image though.

But with that I created through CanDO an email client for the visually impaired as the Amiga could run on a large TV, and also demonstrated how Wordworth could email documents with a simple ARexx script, and I can't remember the spreadsheet application name, but it could check an inbox for the latest email which had an attachment with some data which it then inserted into the spreadsheet, created a graph and then printed it.

Sadly I never released the application (it was my A Level computing project) it was coded in C and the source code is long lost.

The other peeps at college were quite struck by my A3000 which I demoed it all on. Best Amiga I ever had that... (it was pretty vanilla too, just a bit of extra memory, and a Goldengate 386SX bridgeboard, perfect machine really.)

That was the power of ARexx, I loved it!

[turrican3]

Originally Posted by Daedalus

CDDA is a little more complicated, and involves a totally different approach of sending and receiving SCSI commands from the device driver that handles the CD-ROM. This is easy enough to do with system-friendly applications and games, but when the system is shut down, it's made a lot more complicated, potentially even needing different code depending on the interface used.

CD32 versions on WHDLoad are typically hardcoded for the CD32's device driver, which is different to other device drivers found on other Amigas. Also, the nature of WHDLoad is to run the games without requiring the original media, so typically the CD isn't present on the system (and often the system doesn't even have a CD drive). Streaming the audio from the hard drive is even more complicated again when you're dealing with non-system-friendly applications, and the amount of uncompressed data involved would mean a significant proportion of the Amiga's bus time would be used up streaming the audio, meaning big slowdowns in a lot of games with that approach using the standard built-in interfaces.

I know that it's not easy but perhaps more easy than porting a megacd game like another world 2 " Heart of the Alien " and the result should be interesting : new cd32 versions with full cdda support (earok?), they can be used next with cd32 emulators on the a1200 or 4000 with "SimCD32" or

ISO-Boot which is has SimCD32 and the squirrel Power Computing CD32 emu on a boot disk (source : http://eab.abime.net/showthread.php?t=47769) Iso-boot is available on the eab ftp...
It should lead to full 4 voices games for the sfx and cdda for the music.
some exemples : unreal with full 1mb sfx support + cdda music or like i said ninja warriors with in game music, there is plenty possibilities or great games with poor music for which we could switch those bad musics with c64 ones or arcade ones, i think it should worth the pain, and we have a bunch of good amiga music remixes available. thanks for all great amiga music remixers and composers. The biggest problem is the difficulty to make a standard patch like you said, but who knows... i will have a look with the generic slave from jean françois fabre + simcd32 + an iso mount utility but i don't know if one support cdda ??? Or perhaps with some interrupts with nofilecache option and no preload (whdload) and access to cdda directly like when you save on a real disk with whdload but like music normally are launched between stages it should the interrupts shouldn't be a big problem. I will try this crazy way, i'll see what is possible !

[Daedalus]

Yeah, if you're using a CD32 emulator, CDDA should just work provided the standard CD32 access methods are followed. Streaming CDDA from hard drive is doable, but it would surprise you to know just how much of the computer's time that takes up on, say, a standard A1200. One minute of music takes around 10MB, so preloading will be tricky too unless you've a lot of RAM, and transferring that much data could take 10 seconds every minute. So it's not really an option except on much more powerful machines.

[Marle]

I have a question, AGA sprites.

I understand OCS/ECS sprites well enough but not very clear on some of the AGA stuff.

So I know you can have 16x16, 32x32 and 64x64 sprites with AGA, but does that mean all 8 of them have to be the same size, can you mix and match, or is it two sizes on screen at a time e.g. like the SNES which can have 16x16 and 32x32 on screen, but not 16x16, 32x32 and 8x8?

Do the larger sprites occupy 1 or more sprite channels?

Colour wise, do they work like OCS/ECS sprites, 4 colours by default, 16 if attached? Or can they have more colours? I am aware you can shift the colour register offset where sprites take their colours away from col reg 16.

Just curious!