Guaging interest in a Zorro II

[mongo]

So, proper clockport NIC, is it possible?

I'm working on one right now, actually.

[mboehmer_e3b]

Well, it's too easy to fall into feature crawl and overengineering, and then things are delayed, get expensive etc.

It was not overengineering, but not knowing exactly how complicated Zorro III DMA can get if you want to workaround everthing.
From a certain point in time on it was clear that more flexible logic and internal memory was needed, so the CPLD solution was simply not fitting any more.

So, proper clockport NIC, is it possible?

Possible - yes. Nowadays there are plenty of NICs availablem, some with more intelligence, some with less.
Depends on how many interested people are there, and especially who can do driver programming.

It is really hard to estimate on how many modules one can sell, and if the amount of time, work and money put in will ever pay.

Michael

[Boot_WB]

Hi Michael,

(OT I know, but) has there been any progress on the Prometheus firmware update?

Last time I knew (which was around a year ago, and I've missed a lot since then) it was finished, but still some work to do on archives & installers.

I've looked on your website, but couldn't find any news.

Best Regards



Rich

[kolla]

It was not overengineering, but not knowing exactly how complicated Zorro III DMA can get if you want to workaround everthing.

Oh, my comment was not aimed at you, just a generel observation, sorry

[kolla]

I'm working on one right now, actually.

Very cool! As on open design, or with the goal of having a product to sell? Or perhaps both?

[Jose]

Why stick at just ZIII DMA, let's make the impossible and make it ZIII DMA with burst mode.

Just kidding...

@mboehmer_e3b
"So all Zorro II machines are out, all A4000T machines, all A4000 machines with Buster 9 and one DMA Zorro III card, and all A4000x with Buster 11."

Wasn't it you that said it was only Buster11 mahchines with already a ZIII DMA card that were ruled out ?
The 4000T's internal SCSI controler can be disabled...

[pan1k]

Maybe instead of re-inventing the wheel, someone should try and finish up the driver(s) for the RR-Net? It would be funny to have a RR-Net hanging off my X-Surf

[aperez]

Nowadays there are plenty of NICs availablem, some with more intelligence, some with less.
Depends on how many interested people are there, and especially who can do driver programming.

Are there NICs available? Yes. Are they overpriced? Absolutely. Zorro NICs are routinely selling for well over $100-$150 USD and there are *not* functioning Ethernet options for the A500 or A1000.

Agreed, but I am not subject to the regulartory issues you are in Germany, and this is why these sorts of projects are better suited to being open, given the age of the platform. Cheap runs are still viable.

[Tension]

About SATA: SATA controller chips are usually only PCI based. And they need DMA to operate. So all Zorro II machines are out, all A4000T machines, all A4000 machines with Buster 9 and one DMA Zorro III card, and all A4000x with Buster 11.

So pretty soon we are all screwed!!

Someone will make it work... hopefully!!

[aperez]

The obvious solution is to bring native, non-bridged PCI to the Amiga. Zorro is dead, and ZIII-capable Super Busters cost upwards of $40.

[Kronos]

@aperez
And where do you want to connect those PCI-slots to if not Zorro ?

The A3/4000 CPU-slot ? Hard to find parts mechanical nightmare (asyou ould have to provide ways to connect an accel and place the PCI-cards in a sensible way.

A1200 Trapdoor ? Same as above plus being limited to 24Bit address-space.

Directly under the CPU ? Might work for a one-off prototyp ....

The BV/CV-PPC slots on the Phase5 cards ? Guess what someone allready thought of that.

[aperez]

I'm well aware of the limitations you've outlined, and I think the best way to work around the challenges they present is to do it *where the bus arbitration actually happens* eg, Super Buster. You've got a standard 84-pin PLCC socket to work with. 84-pin, 5V high-density CPLDs are *much* cheaper and easier to obtain. The basic logic needed to bring the machine into a booted/usable state can be captured with the logic analyzer I already possess.

[matthey]

@aperez
Are you working on a replacement Buster with PCI expansion using CPLDs? I wondered why CPLDs or FPGAs had not been used for such an expansion/Buster replacement already. I was thinking of a small riser board above the Buster with a cable running to the PCI expansion slots. If you drop a CPLD in the Buster socket, maybe you are thinking of still plugging into the Zorro slots like the Mediator or Prometheus?

[RMK305]

$75 for a zorro ethernet card? I'd buy one.

[aperez]

Maybe instead of re-inventing the wheel, someone should try and finish up the driver(s) for the RR-Net? It would be funny to have a RR-Net hanging off my X-Surf

Pan,

There's nothing to "finish" since this oft-spoken-of driver is not available in any form, much less source. What would you propose I do about that?

The CS8900A which the RR-Net is based on is available on a DIP-socketed proto board from people like SparkFun for $35:

Additionally, the CS8900A in the RR-Net is used in 8-bit mode, which Jens himself has stated does not allow the use of interrupts: "The chip is used in 8-bit mode, so the 8 registers of an NE2000 are spread over 16 registers in an 8-bit system. The 8-bit mode of the chip does not support IRQs (see Cirrus logic application note AN181)."

That said, the chip itself does support interrupt-mode, and Jens' documentation states that "Although the Amiga makes the interrupt status queue registers available, it does not have any effect. Even if you try to activate the chip's IRQ features, it will not have any effect. The IRQ line of the chip is not wired on RR-Net at all!" but the pin for it is not passed through on the overpriced RR-Net.

And, taken directly from the Cirrus Logic Application Note itself:
"Unsupported functions in 8 bit mode:

* Interrupts are not supported. Polled mode must be used
* The DMA engine only uses 16-bit memory accesses and does not support 8 bit transfers.
* The packet page pointer has an auto increment feature that cannot be used in 8-bit mode
* An EEPROM  is not supported"

Citations:
http://www.cirrus.com/jp/pubs/appNote/an181.pdf
From: http://www.schoenfeld.de/inside/Inside_RRnet.txt

http://www.sparkfun.com/commerce/product_info.php?products_id=200