The MP3 file type has been around for so long, and is supported by essentially all modern media software and hardware, that it might be surprising to some to learn that it’s actually a proprietary format. Developed in the late 80s and early 90s, it rose to prominence during the Napster/Limewire era of the early 00s and became the de facto standard for digital music, but not all computers in these eras could play this filetype. This includes the Amigas of the early 90s, with one rare exception: this unreleased successor to the A3000 with a DSP chip, which now also has the software to play back these digital tunes.
The AA3000, developed as a prototype by Commodore, was never released to the general public. Unlike the original A3000 this one would have included a digital signal processing chip from AT&T called the DSP3210 which would have greatly enhanced its audio capabilities. A few prototype boards did make it out into the hands of the public, and the retrocomputing scene has used them to develop replicas of these rare machines. [Wrangler] used one to then develop the software needed for the MPEG layer 2 and 3 decoder using this extra hardware, since the original Amiga 3000 was not powerful enough on its own to play these files back.
If you want to follow along with the community still developing for this platform there’s a form post with some more detail for this specific build (although you may need to translate from German). [Wrangler] additionally points out that there are some limitations with this implementation as well, so you likely won’t get Winamp-level performance with this system, but for the Amiga fans out there it’s an excellent expansion of this computer’s capabilities nonetheless.
Thanks to [Andy] for the tip!
I think you mean “Commodore”. “Amiga” was the computer line and the original company purchased by Commodore before the product was released to the public.
I still have my original Amiga “B”2000 and A3000 computers. It’s a shame Commodore wasn’t able to make it successful long-living product, but as frequently reported, management was more interested in lining their pockets than running a successful company.
In the early 1980s Jay Miner, along with other Atari staffers, became fed up with management and left the company.
In September 1982,[2] they set up another chipset project under a new company in Santa Clara, California, called Hi-Toro (which meant “high bull” to them,[3] later renamed to Amiga), where they could have some creative freedom.
As I noted, this was the original company that was purchased by Commodore. The “AA3000” was prototyped by Commodore long after, not by the original Amiga Corp. that no longer existed.
The group I was in when I developed this machine, at Commodore in West Chester, was at various times shown as Commodore Business Machines, Commodore Semiconductor Group, Commodore Technology, and yes, Commodore Amiga. It was in 1991, years after Commodore had bought the Amiga company. That company was integrated into Commodore in various ways and the technology did keep evolving, despite major flaws in the management of the company.
Hearing from original development staff, this is awesome, thanks. It’d be great to hear more from these kinds of perspectives.
oh, yeah this is in response to whatever I say. Irony, I would call them All At Commodore,as I worked with Jay since 1974. Jay Miner Never Left Amiga .Why, he Owned Everything Then Gave it to I. Everyone knows Knows this. The AA or 24bit without ham came out in 1988. All, Amigas Coulg play All Mpegs. In fact there was no standard. The CD32 could play All formats Including the Pioneer Redray ..granted a little adjusting. Then, it ..the CD32 could play mpeg 18 format. They, had released mpeg 6, but the Amiga Toaster Killed,and Still is the best. Could go on, either most of these people lie or dont know jack.. the ones commenting ,and this so called news article person,and resource. Jay, Dave,and Karl never every disagreed with me,and in fact left the Amiga community ,before Jay Died or what was left it as They, said,and Only went To the Y2K Show in St.Louis as well as All the clebs ,and ceos As I was there, and You All Know it.
RHRH Constantine XII-AmigaMan
I did spend about nine years at Commodore designing exclusively Amiga hardware, like the “B2000” (aka A2000-CR), the A2620/A2630 CPU boards, the Amiga 3000, the A3000+/AA3000, Nyx (AAA chip development system), etc. I think “Commodore Amiga” is appropriate.
You are a legend! Listened to a few of your interviews on YouTube recently, must have sucked working hard on amazing projects like the A3000 and then Commodore making such awful business decisions.
May I second Richard’s comment? As an avid Amiga user and (hobbyist) developer from circa 1989 onwards (A500 later upgraded to a 2.04 ROM; A1200, A2000, A4000/040… and my original Workbench & drives now live on in emulation thanks to UAE), I recognize your name, Dave and am very, very grateful for all your contributions to the platform. In fact, I think we conversed once or twice on the Amiga usenet groups way, way back in the day and you were far more patient and helpful than a wet-behind-the-ears student had any right to expect…).
I’m very happy that you’re still around & online – and, I hope, thriving.
(tries to keep rabid fanboy gleam out of eyes fails)
LOL. Absolutely, and fixed.
The article says:
And the project page says:
I’m pretty sure that if we play the game of lowering requirements, we could assume that the original Amiga 3000 was powerful enough to play MP3 files, as long as we’re talking about 8 kbps mono ;-)
And if we increase system requirements, an m68040 would play a normal 128 kbps MP3, and an m68060 would play any kind of MP3.
I never thought to try that on my 040 Amiga 1200. Now I will!
Hopefully 33 MHz is enough as 40 MHz isn’t quite stable for me.
I’m pretty sure that I was able to play 128kbps MP3s on my A4000/040, albeit Tracker/MED/OctaMED tunes were more prevalent at the time. If I get time later today I’ll fire up WinUAE with various machine speed presets and see if my memory’s accurate.
Offline conversion could also work, but would take a lot of storage to get more than 8 kHz mono 8-bit µlaw.
That does seem oddly weak. The DSP3210 could run 32-bit floating point algorithms generally 5-10x faster than a 68040, much less a 68030. This works via internal SRAM, zero overhead loops, MAC instructions — the usual DSP stuff — and a 50MHz CPU clock in the otherwise 25MHz system.
I did see one of these boards but not all the details. The original project was called the Amiga3000+ at Commodore. Along with the DSP, I had a 16-bit Audio CODEC and an audio mixer so that Amiga and DSP audio could play through the same outputs, however you liked, but there was no need to go through Paula to play back audio. If they don’t have direct audio output, maybe some of the restrictions are to better map into the Amiga’s audio, but I’d want that as a transcoder on the DSP, not a file format restriction.
We could play off the shelf 128kb/s MP3 files some years later on the Metabox Phoenix set top box prototypes. I ran a Motorola/Freescale ColdFire at 90MHz and 144MHz depending on model, with SDRAM. The ColdFire did offer MAC instructions, but no hardware floating point. On the 90MHz version that did chew up something like 90% of the CPU.
The DSP3210 by AT&T was also used by the Apple Mac AV series of computers to power their video and audio capabilities before the PowerPC Macs were introduced. The PowerPC was powerful enough that the DSP was not needed for the AV capabilities.
and plenty of Mac Photoshop accelerators, usually in DSP pairs
Apple got a big boost on their AV project based on our work at Commodore on the A3000+/AA3000. Jeff Porter and I visited AT&T in 1990 to talk about our plans for the DSP3210. Their usual business model was selling DSPs as a replacement for big boxes of analog hardware. You’d buy the chips and license the software as you needed it for your specific job. We had a different plan: we were adding the DSP3210 as a general computing resource, and we wanted all the software libraries to be available for general use by application software. We finally got the AT&T people to understand what we were after, and got a low cost blanket license for nearly everything… Not the 9600 baud modem, but we’d have the 2400 baud modem in the box.
So of course the management idiocy at Commodore ensured the machine never came out. In fact, the new management team had pronounced it a firing offense to put a AA3000 main board in an A3000 case… Good thing Bill Sydnes never visited my garage! But when Apple later went to AT&T, that kind of use was already in place.
Apple’s AV machines never worked as well as our system would have. The DSP ran AT&T’s multitasking VCOS/VCAS operating system, which was a near perfect match to AmigaOS.
Did you, by any chance, at AT&T meet George Warner or Kreg Ulery? The former was the contact I found in some ancient README and tracked down on the net. He held the key to get the DSP software development restarted for us.
The reason for this was that floating around on the net were only parts of the DSP manual. George was incredibly helpful. He not only scraped together all DSP software and development bits he still found on ancient hard drives, he also drove all the way down to Michigan from Pennsylvania to meet his former boss at AT&T, Kreg Ulery, who happened to have written the DSP manual and kept a copy (perhaps the last one in existance) in his attic he scanned for us.
They both expressed their fondness of this particular chip and were very happy to help.
“In fact, the new management team had pronounced it a firing offense to put a AA3000 main board in an A3000 case…”
Wow. That is ridiculous and petty and short sighted and… sadly, very much “on brand” for what I read & observed of the CBM management team in the latter years. Truly, to borrow the military quote, “Lions led by donkeys”.
So not as powerful as the 56k in the falcon which could just about manage a 320kbit mp3 in the lowest resolution that allowed to mp3 player to be usable. Amazing work and really interesting stuff I never even knew this was planned for the Amiga lineup!
The DSP3210 was significantly more powerful than any 56K processor… ours could peak at around 25MFLOPS. It multitasked and did 32-bit floating point 5x-10x faster than the 68040. It also shared the main bus with the CPU (68030 or 68040), so all dram could be reached directly, not sitting out on an I/O port like the MC56K. So they have some limitations that’s not clear… Probably formatting audio to play native on the Amiga legacy hardware. I’ll probably see these guys at Amiga40 in Dusseldorf this year… I’ll want the whole story. Could be this is just the first prototype and they’ll improve it later.
The original Amiga3000+ at Commodore has its own 16-bit Audio CODEC that could mix with legacy Paula audio… No need to transcode for playback. Just a guess. The AA3000 design these are based on was after the new Commodore management forced me to remove all the extra audio stuff in the prototypes, but I kept the improved DSP design. Sadly, a bug in the interface hardware added to the last version of the DMAC chip meant I had to hack in the interface with some dead but PALs. So we only had three DSP machines of the 50 or so AA3000 boards.
oh, hey And I dont get the AA ,and Hombre Confused Anymore.lol..Thanks for the Long Talks of Knowledge Since ,The Early 80s. Yes, Cant Forget Karl ..we need rebol as js is pig..As, you know I open sourced All Amiga as Jay wanted granted it was tech in 91.Yes, I miss those long talks with Jay, who doesnt. The, best one was the first one in 74 as ,Always…the unexpected is Always the best. Shawn Constantine XII –
Just get an Amiga 486 bridgeboard, Sound Blaster/WSS card and run WinPlay3 on Windows 3.1! :)
My 68040 played them well enough… Never really thought to much about it, it worked.
I used to play 128kbps mp3s on my A1200 with ‘030 accelerator board and FPU back in the very late 90’s/early 2000’s, as were many others at the time. It couldn’t handle much above that bitrate very well but considering I still had dialup at the time it didn’t matter, I could still get my Weird Al on just the same.
Apparently 4 MP3 players that work with released Amiga hardware and a 68040 processor or faster…
https://aminet.net/search?query=mp3+player