New A-EON Music Card for Classic Amigas

[spirantho]

A 7MHz A1500 can throw data across the bus fast enough to keep it fed with 320Kbps mp3 data, so I reckon a 68020 over the clockport should be similar.

[spirantho]

It's something we're looking into, definitely. I've not looked at the mpeg library for a while so can't remember the API.. but it would be nice to be widely supported by existing software as well as the new programs!

[spirantho]

From the press release it sounds like it will only do 8-bit when connected via the clock port.

The data bus on the clock port is 8-bit, but the card does the dirty work decoding the data - and however you're sending your OGG Vorbis file or MP3 of whatever (or even WAV) it makes no difference to the result as the supplied data is the same - it just takes twice as many writes across the bus (as it's 8 bits at a time instead of 16).

@nicholas

Access is done via a library but I'm not sure of how open it's going to be (though I'm making it as easy to use as possible). That's an AmigaKit decision....

@Matt_H

Thanks, I'll take a look!

[spirantho]

I wonder if it will be possible to use this to encode to MP3 the audio coming from Paula, as encoding is one of its major features.

The data transfer width (16 for Zorro, 8 for clockport) will merely affect the max bitrate that can be fed into the device.  I have no idea what the A1200 clockport can achieve, bandwidth-wise (1.75MB/s is suggested at http://www.amiga.org/forums/showthread.php?t=24079 - that's enough for several MP3 streams).

I reckon the clockport has enough bandwidth. I'm going to be attempting it fairly soon, but I reckon if a 68000 can pump data fast enough on Z-II, the 68020 on the clockport should cope even with an 8-bit bus. Not sure about a 68000 with a clockport, but there's no harm in trying.

As for your first question - the priority of this card is decoding rather than encoding, but yes - the hardware is capable of encoding Paula into OGGs and MP3s and whatevers.

Edit: Incidentally, I think it goes without saying that if the people behind this card (and that includes myself) do well from it, then the chances of future hardware on classic Amigas is much higher than if we do badly.

[spirantho]

@nicholas
Everybody should be using OGG Vorbis files anyway. Completely free, open, smaller and better sounding. Kind of a win-win-win-win situation, really!

For playing music files this will win hands-down over the Deneb sound card. I have one on my A4000 and although it works it takes a lot of CPU time, even on my CS-PPC 060. The Prisma's only workload is feeding the chip the compressed audio stream.

[spirantho]

The price hasn't been decided yet, but it won't be ridiculous. Unlikely to be £50, though, I'm afraid, from experience. By the time you've factored in manufacturing of the board, the parts, the transport costs, the software costs, the hardware design costs and of course taxes, and with the fact that this is going to be relatively low numbers in the production run, £50 would leave practically nothing as profit for the company (they have to make some, even Amiga hardware makers have to eat ).

What I can say is that we're aiming for a good price for everybody, including the buyer. No-one's going to get massively rich off this one board (and if they are, I want a bigger cut )

[spirantho]

Hope it offers something beyond what the used market provides.



I already have a Delfina Plus, that only uses 2% cpu to decode, has great sound quality, and is bullet proof in reliability.  My 3k multitask without any perceived slowdown now. If this will offer something better, I'll entertain it (I will probably buy if it as least as good as what I have now for my spare desktop, but it would be nice to offer more).  

The Delfina Plus only does MP3 doesn't it? This does OGG Vorbis, WMA, WAV, FLAC and MPEG 2 Layer 3.

The problem, as has always been in Amiga Land, is software and driver development. Hopefully, they will have a good and well tested solution out of the gate.

I'll do my best.

[spirantho]

I believe the clock-port on the Prisma is either host or slave, but not both at the same time. That said, it may be possible to do something funky, not sure yet... (but we do need to keep an eye on costs as well)

[spirantho]

Edit: Double-post. Silly forum. First the post disappears completely, then I re-post it and it appears twice.

[spirantho]

Don't worry, I was being facetious. I know there's good reasons for using mp3.
Well, one good reason anyway.... The one you said.


Mpeg 2 layer 3 may also be .mp3 but it doesn't mean all mp3 playing devices support it.

If it's all right with Matthew, then anyone wanting software features is free to email me about them, but I can't promise I'll be able to implement them!

[spirantho]

After read somethings seems like a MASplayer for clockport.

Imagine a MAS player that provides a click port or can be connected via one.
Then imagine that it can play OGG,  FLAC, WMA and WAV.
Then imagine it can encode audio input such as Paula into OGG or MP3 or whatever.
Then you have the Prisma.

[spirantho]

What about AHI support?

Are you the guy doing the software library for this device?

Yes, I'm "the guy"

Feel free to email me with ideas, suggestions and requests.

AHI is a possibility but not definite yet as the hardware is more geared to decoding than raw playing- but does support raw wav playback.

[spirantho]

AHI is something we're looking at, definitely.
Datatypes aren't possible though, I'm fairly sure. An audio datatype has to convert its audio to an 8svx block of data, which the sound datatype then processes and plays back itself. The Prisma card plays the sound itself, it never converts the data, so there's no data to pass to the audio datatype superclass. I thought of faking the audio by playing the stream and  just giving it zeroes to play, but the datatype doesn't have control of its playback once it's returned its data so it'd be a bit useless.

[spirantho]

That's what I thought. It'd be a lot of work (especially to make sure there were no unwanted repercussions from hacking a major datatype) and I don't think it'd be worth it.
The only real advantage to using a datatype would be to allow programs like Multiview to run it. Datatypes are useful in some apps for importing data, but that's not an option here because there's no data to import (it just gets played directly). APIs like SDL shouldn't be using datatypes anyway unless it's for importing data - it's a rather inefficient way to actually play data, and it offers no control to the calling application.

If I manage to get the AHI sound driver working (it's not retargetable audio, remember - it's actually a stream player) then of course you could use the mp3 datatype to decode/import and then play it through AHI, but I think it'd be rather easier to just use the chip to play the mp3.

[spirantho]

so the almost only advantage of this card is that you will be to play your audio collection stored in some compressed format using a dedicated player? at least something like mpga.library should be thought of. or are we facing c64 future wehre everything hast to be statically linked?

*shudder*
I hate static linking. What do you think I am, a Linux coder?

Here's what you currently need to do to play an OGG Vorbis file:

PrismaBase = OpenLibrary( "prisma.library", 0L);
Prisma_Init();
Prisma_PlayFile( "GreatMusic.ogg" );
CloseLibrary( PrismaBase);

There are a few extra parameters to PlayFile if you want to use them (volume, balance, buffersize, task priority), but that's pretty much it.
There's also PlayMemory() which I'm going to implement soon, and it's this that'll hook into AHI and any apps such as SCUMM.