@PPC CPU for classics

(my summarypost)
"538s blender 2.49 MCP8610 1066Mhz"

Also interesting for HW builders, even if not "new" chip.

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

@pavlor

Quote:

Their project bears name "UltimatePPC" - that means OS4, or bust.

not entirely.. they tried to convince the warpos designer in, apparently some changes have to be made since then. however if it really means only os4, then chances are rather for bust, i sense.

@Rob

Quote:

Rob wrote: @DBAlex Quote: Didn't they demo a prototype SharkPPC running OS3.9 a couple of years back? Possibly thinking of different project but was sure I've seen it somewhere! You're thinking of their Dragon board. I guess they didn't get the performance they were looking for with coldfire. Not sure why they didn't make a 68k CPU board for the Dragon since it would have been a much better setup than the Mediator.

Yes you are correct. Was thinking of the Coldfire project. Still, quite interesting... Haven't checked progress on that for a while. Probably gone the same way as SharkPPC.

(First post from BlackBerry PlayBook! Very impressed so far, especially with new OS update)

_________________
A1200, 68060/64MB/1.2GB/WiFi/AGAtoCRT/OS3.9 Pegasos I, G3 600Mhz/512/9200SE/80GB WinUAE, Ryzen 5 2400G/Vega11, 8GB DDR4, 256GB SSD,Win 10 Pro x64 Amiga Forever !

do people still want PPC for classic Amiga's?

I just assumed when Amiga inc announced Os4 and that all the licensed computers running it would be referred to as NG Amiga's we would all slowly move in that direction.

_________________

@Panthro

Sadly Amiga Inc sold out to various companies, including Commodore USA and the only modern computers that bear the Amiga are PCs.

@Panthro

>do people still want PPC for classic Amiga's?

I will definitly buy several, but no one after phase 5 has the knowlidge so far to create them.

Last edited by bennymee on 27-Dec-2012 at 08:24 AM.

@bennymee

One of the reasons of classic PPC progress slowdown was that Phase5 controlled both HW and SW. And single CPU PPC cards were not really possible (by other third parties) untill AOS was ported to PPC (remembering that during those old times MOS was phase5 guys OS). And the biggest slowdown during this millennium was because of AmigaInc.

Today HW design is easier (SoC chips, FPGAs everywhere, etc) and OSs are in the hands of active developers.

So, not the biggest surprice in the world if we one day see a new & working PPC CPU card.

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

@persia

Well they were the ones that came up with Amiga NG and the resultant AmigaOne naming as well
although I cant send you a link as Amiga.Inc got rid of that quick smart when the Amiga OS4 court battles hit.

_________________

Individual computers have a strange strategy..Producing all kind of stuff, and accelerators but no PPC..similar to Elbox
Is it so complicated technology..
How about to by rights from b-plan and produce it with modern proccesors some usb ports, more memory ..?

_________________
F1 Srbija

@Comi

Quote:

Is it so complicated technology..

Yes, it is. There is much greater difference between plain 68030 (with proven software compatibility) and PowerPC. What about 68k OS3 compatibility, or do you want OS4 only CPU card? Blizzard/CyberstormPPC offer both full 68k and PowerPC compatibility, but these boards are too complex to be re-created.

@Comi

Quote:

Individual computers have a strange strategy..

no wonder since they consider the whole ppc story a failure from the start on.

@wawa

So they believe that new hardware should be based on....which processors coldfire ?

_________________
F1 Srbija

@pavlor

One possible problem is that the vendor must buy a certain amount of CPUs in order to have them at "reasonable" price.

My own experiance with bluetooth chips:
The idea was to develop a bluetooth protocol for A-like Oses.
So it seemed to me the easier was to secure a hundred of those bluetooth chipsets (in a usb stick) in order to provide them to customers (why? because it's almost impossible to know which vendor integrated which chipset from x or y vendor in their product and I wanted to ensure full compatibility and not redeveloping drivers eternally).

So I asked to texas instruments for some prices.
In order to have them less expensive that any bluetooth stick available in any supermarket, I was asked by Texas Instrument to order 10.000 units, something I haven't the money to, and would not sell 10.000 units in the little amiga market
I had in mind a batch of a hundred of those sticks, I could get them *3 more expensive than the counter part (same chipset) in a little supermaket 1km away from my home, then on those I should apply a margin, as business is business (even if it's a tiny business).

So I asked broadcom. for a 100x batch, I was asked 2x time more money than the ones in the supermarket. There again, I had to order some 10.000 units of those.

The only solution I see there is to ask a distributor how many it costs for this bluetooth sticks, and there again, the prices are too high, because they are not interested to sell all these to a single man at the same price they apply for supermaket back office buyers....

So well, coming back to the PPC cards projects, you can have a PPC card proto, but when it comes to order a batch of some dozens of those cpus, or eventually a hundred, you see that scale costs rules a lot!

If it is to sell a PPC card at a price of 1000$, while you were targetting a max of 500$ you know it certainly doesn't worth investigating more....

I wonder how current amiga hardware vendors are dealing with this situation that brings headhaches for little projects directors!

Last edited by DC_Edge on 27-Dec-2012 at 12:35 PM.

@DC_Edge

When there is no competition, Amiga HW manufacturers can buy smaller patches and price the products accordingly.

From my very little study, it seems one could build 1Ghz PPC Amiga with about 100eur parts, without the need for large patch.
Paying for HW R&D, OS porting and HW product testing is more challenging.

(but with 100eur cost for the HW, its easier to arrange 10 000 unit build than with x1000 or SAM (just as an example)kind of HW)

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

@Comi


here is a direct link to jens' statement on a german forum. use some online translation if you will. ill yust quote the beginning: "my opinion was from the very start, that ppc was the worst that could happen to the amiga..."
http://www.a1k.org/forum/showthread.php?p=445998&highlight=ppc#post445998
(post #387)

Quote:

So they believe that new hardware should be based on....which processors coldfire ?

i dont know what jens opinions on this exactly is, he rarely lets it drop through, so dont quote it on me, but its likely an amithlon like solution would be of his prefference. when x86/68k emu accelerator idea for amigas was mentioned to him lately he seemed to think that it would be doable and might make sense. other than he is obviously hardcore 68k.

on the other hand his opinion about the guys behind ultimateppc is very positive.

I read russian wikpedia about Freescale Coldfire and it says:

"The latest models of Coldfire have become quite compatible with 68k to run Amiga and Atari ST code"

I don't know who wrote this, but I also surfed web and found that latest versions of Coldfire (v4 and v5) is more compatible to original 68k than earlier version. So... what do we waiting for? :)


Here: http://www.microapl.co.uk/Porting/ColdFire/cf_68k_diffs.html

"Version 4 of the ColdFire core architecture re-introduces some familiar 680x0 instructions, and also adds some new instructions. The main changes are .... "


And there is a part of the text about Coldfire v5:

"At 333 MHz and 610 million instructions per second (MIPS) in 0.13 micron process technology, the V5 core offers significant performance increases over the V4 ColdFire core, previously the highest performing ColdFire core on the market.
"

So may be 610 MIPS is enough for proper 68k emulation?

Last edited by coriolis on 27-Dec-2012 at 04:31 PM.
Last edited by coriolis on 27-Dec-2012 at 04:29 PM.

_________________
A1200 + (looking for bppc060) + bvppc + Indivision AGA MKII + OS 3.9
A3000 + OS 3.9

@coriolis

I don´t think Wikipedia is best source for such kind of informations.

FireBee project (ColdFire based motherboard for TOS/MiNT) could show us, if Coldfire is good idea for our needs, or not.

However, ColdFire accellerator for A1200 without integrated GFX card is pointless.

@coriolis

Quote:

"Version 4 of the ColdFire core architecture re-introduces some familiar 680x0 instructions, and also adds some new instructions. The main changes are .... "

has been known and discussed ages and ages..

Quote:

what do we waiting for?

well, what are *you* waiting for?

@wawa

I'm not an electronics-guru. But I remember one guy who also wanted to make a CF-accelerator for A1200 - http://turbo-cf.narod.ru/eng/index.html

Here are a some schematics - http://turbo-cf.narod.ru/eng/download.html of this project. May be it will be helpful for somebody?

_________________
A1200 + (looking for bppc060) + bvppc + Indivision AGA MKII + OS 3.9
A3000 + OS 3.9

Quote:

coriolis wrote: I read russian wikpedia about Freescale Coldfire and it says: "The latest models of Coldfire have become quite compatible with 68k to run Amiga and Atari ST code" I don't know who wrote this, but I also surfed web and found that latest versions of Coldfire (v4 and v5) is more compatible to original 68k than earlier version. So... what do we waiting for? :)

The ColdFire and 68k are incompatible because some instructions with the same encoding have significantly different behavior. This means that 68k code can't be simply trapped on the ColdFire. Code must be patched before it's executed which is slow and less compatible. Some of these incompatibilities are common too. For example, the stack on the 68k is word aligned where it's longword aligned on the ColdFire. A 68k fpga CPU or full software emulation of the 68k is a better solution IMO.