@KimmoK

I wouldn't go that far, first the design should be made. I've had an idea about a bounty scheme that might work, I'll post it in a while.

@feanor

I am still interested in the idea. Open sourcing the hardware design might be a good thing, but OTOH it will be difficult to source all required parts to do minimal prodction runs. Anyway, open sourcing the design could benefit a lot because it opens the design to new target groups. Maybe some tech faculty of this or that university may get included to do some work for educational reasons.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)

@Zylesea

Quote:

I am still interested in the idea. Open sourcing the hardware design might be a good thing,

IMR the PegasosII was open sourced ...

_________________
AmigaONE. Haha. Just because you can put label on it does not make it Amiga.

I borrowed this comments from here (#27 & #28):
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=38873&forum=2&start=20&order=0

@Arko
Quote:

IMR the PegasosII was open sourced ...

It is. However, it's not RoHS compliant, which means it needs extra design cost to make it RoHS compliant -otherwise it's illegal to distribute it in the EU (possibly in the US as well?). Also, the components right now are too expensive, and it would cost to much to get for low specs (would you be ok, with AGP only and USB1?), it's actually better/cheaper in the long run to design a new motherboard, based on new specs.

Last edited by feanor on 11-Nov-2009 at 01:23 PM.

Um, this is an insane idea from me, the insane idea person, but how powerful are various FPGAs, for AMIGA OS use, that is, lower speed needed through effiicency etc, if you could use Open Soft Cores on them, for PCI Express, USB 2, SATA, System Support CoProcessors etc?

The more I see of Mobos, present designs, the Less the external IO gets, in using serial point to point connections with pairs of wires, and DDR,2,3 Dual, Tripple channel, with ever increasing Parrallel busses and vastly increased real estate needed in absolute terms, and even more in relative terms.

Is the Mezzanine, CPU card any good, if its used Especially for the single reason not of supporting CPU upgrades and interfaces on a given long life board, although that would be a nice side effect, but in that the Memory is specific to the CPU card? Literally go for the Classic Amiga style of Dedicated CPU Fast RAM, and System RAM, Graphics CArd RAM, Cache RAM on various devices, so you have a flat memory model, but part of the Address is actually local system layout?

Just thinking of MAC address at 128 bit, and hardware UUIDs ?? being 64 bit, and IPv6 is 128 bit?.. IPv6=MAC?? So couldnt a 256 bit flat memory model be possible, with auto extension to 1024 plus to bypass the needs for a major rewrite in another 30 years or so?

Something useful, The Iterated Fractal Analysis algorithm patents expire this year?.. Fractals are, after 20 years (patents?) becoming of intrest once more, especailly as they appear absolutely everywhere, and are useful for everything. Like Fourier Transforms. These would be useful pieces of hardware to implement as they are seriously heavy on code, and yet in many cases are very hardware light. Call them next generation FPUs, but only if theyre implemented in complex number format.

But then, how much would such a setup even cost to research and develop, never mind implement? Teh security aspect would be intresting though. Insert USB key with fingerprint scanner, power up, and the computer doesnt even exist until after that point, which means you could implement hardware encryption routines that depend on the system, user, so even the computer doesnt know what its doing?

Sorry 8(.. Im not very good at this, or Id have certain simple bits of code out already.

Congratulations on trying so far. 8)

_________________
The older and more respected a scientist is, the longer it takes to prove him wrong.

@Arko

Quote:

IMR the PegasosII was open sourced ...

For anyone here to benefit from an "open-source" hardware design, it really should include open-sourcing the Gerbers too. I don't remember how detailed it was now, but the BOM should do more than just say capacitor of how many uF, it should give a manufacturer and part number as well. For a long time I didn't consider a BOM to have been included, thinking for example that the CPU "Slot-1" was an instance name, not a connector spec. The schematics released is an interesting view into how schematics are made, but in terms of me turning that into an assembled PCB, it's rather useless. We don't all have pro-level PDB layout software to redo all that part o fhte engineering design, and even if someone could pirate or even buy these expensive tools, how many of us have the education and skills to make it work after assembly? Send the gerbers and build docs to a fab and it you don't have to be a layout engineer to make use of it. I myself was rather disappointed with what Genesi released, previous discussions in a variety of places made it sound like there would be more than just PDF schematics which are not usable in any PCB EDA tools that I'm aware of, an engineer would have to redo all those schematics in their PCB tool before they could begin layout)

_________________
All glory to the Hypnotoad!

@billt

All design files, Gerber files, hardware schematics, will be opened -we haven't decided on the license yet, but it will be open- stay tuned.

@feanor

I think the bounty is a pretty good idea, the open source design will make it a lot less risky for companies like ACube to produce the board (it seems to be working pretty well for MiniMig).

My design ideas:

Simplified SAMflex board. (how much would Acube ask for opensourcing their design?) (BOUNTY1alternativeA)
Lattice Chip wired so that there are places for Atari-style joystick sockets (4..8ports) on the motherboard, for retrogaming and data acquistion. (or separate design for that via the 80 pin connector)

Perhaps there should also be available designs of simple boards to expand the I/O fromn the 80-pin connector. For things like Amiga floppy drives.

Later: SoC upgrade to a >1Ghz chip with L2 and PCI16x.
(how much would Acube ask for upgrading and oipensourcing their design? (bounty1 part2)
How much would David Haynie ask for such a design? (BOUNTY1alternativeB)
How much would the prototyping and testing with linux cost? (bounty1 part3)
How much would the AOS4 port cost? (bounty2))

Last edited by KimmoK on 12-Nov-2009 at 07:46 AM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@thread

Look - whatever happens the product that will work will be based on an existing evaluation system - that is where the A1 series came from, and the SAM from what I can tell. Nobody has the resources to build something from scratch. Not in this market (yet).

This means leverage something that is out there now and reasonably mature. Best looking item at the moment appears to be the MPC8572DS. Tweak it (minimally) to an acceptable level and build that.

Take the existing evaluation system and get OS4 working on that. The hardware works, it has a u-boot port. Once you get OS4 working getting commitment from users to go into low-level production shouldn't be a problem.

I wonder what is the availability of all these:
PowerPC 460
"Introduced in 2006, the 460 cores are similar to the 440 but reach 1.4 GHz, are developed with multi-core applications in mind and have 24 additional digital signal processing (DSP) instructions. The cores are designed to be low-power but high performance and the 464-H90 is expected to draw only 0.53 W at 1 GHz. The 460 core adheres to Power ISA v.2.03 using the Book III-E specification.

PowerPC 460S – A completely synthesized core and can be licensed from IBM or Synopsys for manufacturing on different foundries. 460S can be configured with different amounts of L1 and L2 cache as well as with or without SMP and FPU.[6]
PowerPC 464-H90 - A 90 nm, hard core (not synthesizable), released in 2006, will offer a customizable core for ASICs that can be manufactured with IBM or at manufacturing facilities at Chartered or Samsung.
PowerPC 464FP-H90 – Released in 2007, is a hard core that adds a double precision floating point unit (FPU). "

Notice that:
"[edit] AMCC 460
The PowerPC 460EX and 460GT from AMCC are, despite their name, processors with the 440 core. They are available at 0.6 to 1.2 GHz and have integrated controllers for DDR or DDR2 SDRAM, USB 2.0, PCIe, SATA, and Gigabit Ethernet.
"

http://en.wikipedia.org/wiki/PowerPC_400#PowerPC_460

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK

apart from the AMCC ones, the IBM 460 ones, are *cores* not CPU chips available soon. Availabilty of the chips themselves is not -at least not AFAIK- announced yet.

@feanor

Putting aside legal aspects of that much money (reporting it for taxes), and who is holding that money with whatever guarantees of repayment, where is the benefit of doing this for PPC arch on the business level? If your designing/producing (because just about anyone can design a mobo, having it turned into functional hardware is another issue) a mobo that is in search of a reason to be used, does not make a lot of sense in this economy. Or am I missing something?

_________________
Dammy

@damocles

there are plenty of reasons to use such the specific mobo, and plenty of producs that could be based off it, if it was available. The exact legal details will be answered soon. As to why PPC? Well, my own personal answer is "because I LIKE it", is that a good enough reason? A more business oriented answer is that PowerPC is not a dead chip, it's dying/dead on the desktop, but as a CPU it's not obsolete, far from it. ARM is cool and the rage nowadays, but it's not up to the PowerPC performance yet.

@damocles

Quote:

a mobo that is in search of a reason to be used, does not make a lot of sense in this economy. Or am I missing something?

from Konstantinos on June 10, 2010:

Quote:

Ok, it is with great sadness that I have to announce that Codex.gr, my company, has officially ceased operations as of a few days ago. Actually it had stopped operations for more than a month, but these days the paperwork was finished. One more victim of the recent crisis...

Source

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@number6

Quote:

Ok, it is with great sadness that I have to announce that Codex.gr, my company, has officially ceased operations

This is really sad. Stinking economy!

I wonder how the open ppc MB project is doing?

@Mechanic

¤64,500 bounty. What do you think?

@Rob

Quote:

¤64,500 bounty. What do you think?

Compared to the 4x that has been spent on the X1000 development, thats a bargain!

And it's for a machine that would have been a lot cheaper but not a lot slower.

_________________
Whyzzat?

@minator

Quote:

And it's for a machine that would have been a lot cheaper but not a lot slower.

PA6T promises much better speed than 8640D/8610 (two times faster - in sense Y=2*X - in integer operations and even faster in FPU operations). However, only real benchmarks can prove such promises.

@minator

Quote:

minator wrote: @Rob Quote: ¤64,500 bounty. What do you think? Compared to the 4x that has been spent on the X1000 development, thats a bargain! And it's for a machine that would have been a lot cheaper but not a lot slower.

that only for the schematics.X1000 have go until producction run stage

_________________
No PowerPC, No Fun
Make Amiga Great Again