This is great stuff, congrats :great. A pity I don't have the money now.

Looks like a lot of work has gone into this.

What sort of transfer rates will we seen on the clock port version on a 030 @ 50mhz?

Thanks

Yep - especially the clockport reliability was tricky to reach.

Since I don't have any 50MHz accelerator, I'll quote mass-storage device transfer rates when used with a 40MHz ACA1233 (which is about the same speed as the Blizzard B1230IV): I measured 712KBytes/s with Sysinfo on an A1200 just a few days ago.

On an A600 with ACA620 (16.67MHz) and RapidRoad connected to the riser card of the A604n, you get 730k/s.

If the ACA620 is overclocked to 25MHz, you get 852k/s, and if you have an ACA630 with 25MHz, you get 876k/second.

Note that the clockport of the A1200 is really slow. The clockport of the riser card of the A604n is a little bit faster, which is why you get better performance, even though the CPU is not all that fast.

Jens

Great work Jens!! Looks like an interesting product for my A1200 and A600...

I like to see a video of this product in action inside an A1200 or A600 and see where the cables are.. These video's will probably show up soon :)

Maybe a stupid question but are there some limitations with certain types of storage-devices? Will any mass-storage device work? Usb-sticks? Do I need a self-powered device or use a usb-hub which has it's own power?

I'm not really a video expert - I leave that up to others.

I haven't found any problems with mass storage devices yet. Poseidon is very tolerant to the various "implementations" (read: bad china-bugs) of cheap USB sticks.

Example: I have a gift from one of the Commodore incarnations, a "PET 128MB". That's a very early USB stick that came with a driver CD. I did have success with this stick on an old Win98SE computer with that CD, but on no other computer; neither Win32-versions, nor under Linux. With Poseidon, the "Pet128MB" stick with the Commodore label works fine.

Power-wise, RapidRoad uses a very sturdy USB power switch: 1.5 Amps per port. That's three times the specified power, and that's "per port", so if your power supply can handle it, you can power fairly old 2.5" USB harddrives with RapidRoad. I would still recommend a powered USB hub if you want to connect lots of devices at the same time.

I need to buy a USB sound card, because yesterday night, I got an update to Poseidon that also supports USB audio. So far, it has only been tested on a clockport with just 48kHz 16 Bit AHI replay - recording isn't tested yet. Only two realtime streams are supported, so you can record and playback at the same time, but no webcam during that session.
Trouble: Amiga is not a realtime system. If multitasking is interrupted for more than 7ms, audio will stutter. Some SCSI controllers may actually cause such a stutter, including the A3000 onboard-SCSI controller.

Jens

That's very interesting. Please keep us informed about your tests.

That's really interesting.

I always had the impression that USB sound cards were really usable on a DENEB because of the ZorroIII interface. Clockport's bandwidth is much much slower than ZIII.
Is there any chance that the stutter you experience is because of the limitations of the clockport's bandwidth?
Because if that't the case I can hardly see any way to overcome this.

Bandwidth is not really an issue for audio. We're talking under 200k/second, which is perfectly possible with the clockport and of course with Z2/Z3 interfaces. The trick is to start a number of USB transfers at the same time: With RapidRoad, the driver can start up to 8 realtime-ISO transfers at the same time (to my knowledge, all previous solutions used a maximum of two simultaneous transfers). Poseidon had to be updated in order to support this number of RT-ISO transfers. I'll upload this update to my Wiki sometime this weekend.

Also, don't get me wrong: We're not experiencing stutter on the clockport. It works fine on the clockport at the programmer's site, using a B1260. It has not been tested on the X-Surf-100 yet. That's only because I don't have a USB sound card - yet. It should arrive today.

Still, on slower Amigas, the 7ms minimum time between data chunks to be moved to/from the sound card may be a dealbreaker. If a SCSI DMA transfer is running during that time, Poseidon *may* pick up data too late (or send it too late), resulting in stuttering audio. Note that this is merely a possibility, not an observation. I'm trying to tell everyone that this is not yet tested.

Jens

I've found some info here http://www.amiga.org/forums/showthread.php?t=50506

like this:

I was just wondering why it needs a 68020 or better, is it just CPU power or some instruction set required?

It's the USB stack, it only runs on 020+

It's quite slow on an 020 (better 030/040/060 to be honest anyway)

There is a USB stack for 68000 CPU's but I don't know if it will work with the Rapid, doubt it. I forget the name of the stack though. It's quite basic compared to Poseiden though.

Do you mean the ANAIIS USB Stack?

That sounds like the one!

I'm pretty sure gilloo (the ANAIIS dev, he's a member on here) would love to update it to work with the RapidRoad - that's if he ends buying one! :D

Programming info is open and online since a few months already:
http://wiki.icomp.de/wiki/USB_module

Jens

Thanks for that, it will make life easier for everyone who wants to mess with it on the low level (I already preordered one from RetroAmi).

:great

As a matter of interest, what would be the consequences for running one Internal Port to, say a card reader ....& then plugging into the wrong port on the back plane?

Haven't tried it, but I'd expect that both USB devices would be "gone" for Poseidon. Since power is fused (software-resettable by Poseidon), I don't think that any of the parts would suffer damage. Otherwise I'd recommend to leave the pin header open ;-)

Jens

Thanks for clarifying. Just put the order through....

All the best Jens & Thank You. :great

I've just ordered mine :)