New CPU card for A4000.

[strim]

Hi.

Me and jarob are designing a new CPU card for A4000. Codename: Decelerator 4030. It's a project we're doing primarily to learn how to interface with local bus slot (so we can do something more ambitious in the future). Though if we succeed with this, there will surely be a small batch of these cards produced, in case anyone wants to replace their A3630 with something cooler .

Anyway, here's a current schematics.

The whole documentation will be open source, so that other people can hack on this.

[hese]

Excellent stuff.

[Jope]

Interesting. :-)

[jbenam]

Definitely interesting! I'll be waiting for the Decelerator 4060 with RAM support

[strim]

I'll be waiting for the Decelerator 4060 with RAM support

Ah, the essence of Amiga community, always waiting for something. Considering availability of 68060 CPUs, you might wait forever...

[jbenam]

Originally Posted by strim

Ah, the essence of Amiga community, always waiting for something. Considering availability of 68060 CPUs, you might wait forever...

Just two more weeks, right?

I have one sitting at home waiting for the right accelerator
I think that if a new 060 board will be ever produced it should be sold without the 060. Then it'll be up to the user to source one!

[wawa]

@strim
radoslaw, are you aware of this project here?
http://www.apollo-core.com/bringup/
just in case you would like to communicate with people with a similar bias and exchange experience or coordinate efforts. they have been posting here but currently are rather to catch on amigacoding.de and a1k.org.
http://www.a1k.org/forum/showthread.php?t=46334
http://www.a1k.org/forum/showthread.php?t=46093
the hardware is going to be available for 68000 based amigas first, so perhaps a colaboration on a3k/a4k cpu interface slot might be interesting.

[Zetr0]

Originally Posted by jbenam

Just two more weeks, right?

I have one sitting at home waiting for the right accelerator
I think that if a new 060 board will be ever produced it should be sold without the 060. Then it'll be up to the user to source one!

The only problem with that my friend is that different 060's need different boot strap code to start up. =/

[strim]

Originally Posted by wawa

@strim
radoslaw, are you aware of this project here?
http://www.apollo-core.com/bringup/

Of course, I heard about Apollo core, but first Decelerator will surely be based on a real CPU, to keep the project as simple as possible.

Future revisions might be based on some soft core, provided that the developers of such core would agree to make it open source. One of my primary goals here is to make the whole project open.

[jbenam]

Originally Posted by Zetr0

The only problem with that my friend is that different 060's need different boot strap code to start up. =/

Are you referring to the EC and LCs? Screw them, they're botched CPUs On such an high-end system you would want an FPU anyway (and even more importantly, an MMU)

[JimDrew]

...and RAM connected directly to the CPU.

[strim]

Originally Posted by JimDrew

...and RAM connected directly to the CPU.

The prototype can take two AS6C3216 or two IS66WVE4M16 (look at the schematics...). So depending on the chips used, 8MB or 16MB of Fast RAM will be available on board.

Also, to guys discussing 68060. Decelerator (as name suggests) is not supposed to outrun your CyberStorm. It is meant to be a more modern replacement for the very low end CPU cards (i.e. A3630 or on-board 68030). Processing power is not the primary concern here, we're sticking with 68030 and a small amount of Fast RAM for now.

I know that this won't appeal to everyone, and that some people would only be pleased with dirt-cheap 1GHz 68060, or something like that. However, anyone that ever tried to build a hardware expansion knows, building something as complicated as CPU card is not an easy task. We still lack experience, knowledge, and time (we all have dayjobs after all) so we are taking small steps. And the first small step is getting card 68030 working. While doing this, we'll try to document as much as possible, so that future Amiga hardware hackers won't have to reinvent the wheel. Taking another small step (like more advanced CPU or SDRAM controller) from there will be far easier, that trying to do everything now. If we tried to create a 68060 SDRAM-based CPU card now, there would be a huge possibility of failure, wasted money, and nothing ever surfacing. We have to be realistic.

Although, it doesn't mean we are just recreating A3630 here. We are planning on a few interesting features, that will be unique among currently existing CPU boards, like full sensors suite for monitoring the temperature, voltage, current, ability to change CPU frequency using software, bootable SD card interface (hopefully with DMA), user customisable Flash ROM. They are not all on the schematics yet (first built prototype will be more or less bare bones).

[ajk]

I applaud your realism

[wawa]

and i applaud the open philosophy. hope the experiment will be rewarding.

[jbenam]

I was talking about an hypotetical future (I have no need for a 030 personally, as I've already got an A3640), obviously it's better to build a base first. 030, then someone will upgrade it to 040, and from there the jump to 060!

The community definitely needs more open-source projects like that (here's hoping that the profitteers just downloading the schematics created by someone's else hard work and producing/populating the PCBs and reselling them for massive profit will stay away from it... )

[Zetr0]

@strim

I really look forward to your 030 project, its great to the community developing product for its own needs.

I think this would make a nice little upgrade for the A3000 as well! good luck with your project!


@jbenam

Hello my friend

Originally Posted by jbenam

Are you referring to the EC and LCs? Screw them, they're botched CPUs On such an high-end system you would want an FPU anyway (and even more importantly, an MMU)

I would agree with you on the lack of features the LC (low cost) and the EC (embedded controller) offer. Yet if memory serves me (and it is getting on a bit) but there is some difference with earlier XC and later MC 060's

[strim]

Originally Posted by jbenam

The community definitely needs more open-source projects like that (here's hoping that the profitteers just downloading the schematics created by someone's else hard work and producing/populating the PCBs and reselling them for massive profit will stay away from it... )

I think there won't be much problem. The production will be fairly complex, 4-6 layers PCB, TSSOP and BGA IC packages, profiteer-amateur-ebayer at home probably won't be able to build the card (or the cost will be way too high). I doubt anyone will be able to produce the card cheaper than us. We completely open sourced the Sakura board, didn't see anyone cloning it yet . It's not a gotek where you take the ready made product, upload firmware and resell it...

[Zetr0]

Ahhhhhhhhhhhh you are THAT strim!!!!

I thought I recognized you... Sadly I was ill when you developed the Sakura project - I had hoped to have a chat about adding USB 1.1... I have some work here, sadly I have not had time to develop it further.

A cheap 4MB PCMCIA card with two ports of USB 1.1 would be a serious benefit for the humble A600 / A1200

[strim]

We are thinking of revised Sakura version, with integrated Ethernet, SD and Flash, might as well think of USB. But have to finish Decelerator first, as it is eating up all our resources now .

[jbenam]

Originally Posted by strim

I think there won't be much problem. The production will be fairly complex, 4-6 layers PCB, TSSOP and BGA IC packages, profiteer-amateur-ebayer at home probably won't be able to build the card (or the cost will be way too high). I doubt anyone will be able to produce the card cheaper than us. We completely open sourced the Sakura board, didn't see anyone cloning it yet . It's not a gotek where you take the ready made product, upload firmware and resell it...

Way to go, guys! If someone's to earn to some money from your hard work, that's gotta be you and no-one else Being open-sourced also means someone will be able to do more production runs in the future, even after you guys will have moved on to other projects.

I'll keep my eyes peeled for the dev progress on the Decelerator. Even if I won't buy one, it will be definitely interesting to follow

Don't forget to post the delicious HW pr0n pics as soon as you've got some prototypes