WTB: Cyberstorm PPC

[Iggy]

I
Obviously, board layout isn't the whole story.  There's plenty of programmed devices to consider as well.

It's a shame we can't just get our hands on the info though.  The technical info is useful for just keeping our machines going, even if we can't ever justify designing out the obselecence/non conformity and doing a new production run.


AH

That top part is what slowed me down with a re-implementation of a Freescale evaluation design. But even that can be overcome if you can figure out what it does (not how, what).

The board layers, yes an x-ray could prove useful, but then you have to figure out how to separate the layers. Another, much longer process, would be to test all the leads for continuity and re-create the schematic from the results.

But would all this hard work make sense? I don't know. But there would be the sense of accomplishment in defeating this and gaining the info.

[Iggy]

It would have to be Thomas Dellert this is the ahole who owns "DCE" he is the guy really responsible for the whole Phase5 going under debacle. I know Thomas and Gerald of bplan personally and I knew that it would have had to have been him to say this. (the other person is Ralph Shmidt since he was in phase5 but i know for sure it wasnt him to say some silly thing like that haha)  Whats really sad is even the bplan guys cant talk to Dellert. And they were partnered at one point, but apparently Dellert screwed them. (This is based on conversations I had with them when we with Motorola @ SNDF Dallas)

Wow Mag! That was controversial enough to get a quote on MorphZone.
You're famous.

[Iggy]

Could be done. It was once part of my job many years ago to reverse multilayer boards for testing and design flaws. I think a 6 layer was the most difficult I ran accross. What's the PPC card, more than 6? At one time long ago when I ran many A3000 repairs, I reversed much of the A3000 motherboard. That was kind of cheating though as I could check my work against the schematics that were available. Made popping out 3000 repairs pretty cinchy after that though.

As tough as the card schematics might be, there'd be no way to repeat the firmware without plucking it out of the existing chips. That'd be a trick. All sounds like great fun to me, but like most everyone, I just don't have the spare time I once did for additional projects.

Plaz

I agree. It could be done, but it would be time consuming and tedious.
Can anyone tell me how much programmable logic was used?

Occasionally I get a rise out of pursuing something painful.

[Iggy]

>Dellert did buy out all PuP related designs

Well then that pretty much settles it. Who else could offer the accelerator design for sale?
Could, after this much time,  Dellert sue someone for reverse engineering this board?

[Iggy]

Ok guys here is a couple of notes

1. We do NOT know if this silly 200k number is accurate.. as its romour (pretty strong too)

2. NOBODY is going to be able reverse engineer the phase 5 ppc card, do you even realize how complex it is?

3. Can we now start talking sensibly

Why in the world would we want to do that?
Of course I realize how complex it is. That's part of the challenge.

As far as "NOBODY" well that just makes me want to take it on.

First question. Does anyone have a non-functional one. Eventual it will probably be necessary to sacrifice one or two functional cards (an expensive proposition). But to start with, I'd like a broken one.

Next question. Does anyone want to lend a hand with this? Part-time it could take a couple years. Regardless of who works on it, I want to release the information to the general public.

[Iggy]

Agreed, no one is ever going to get the phase5 documentation, and Ralph Schmidt would never give out his source code.

It would be easier to make a new CPU card with a FPGA 68k core and a PPC chip.

We don't need Ralph's help with this.

And I agree that an FPGA 68K core may be an attractive option. As would a different PPC processor (the Phase5 card uses a relatively slow one - a 603e isn't it?)

[Iggy]

iggy

What a waste of time, of course nobody is going to help u its silly.

Mag, you're just offering opinions. In time I hope to change those. As I've said before, I used to work for a hardware manufacturer. What you think is impossible is going on every day.

I'm going to go dig up the available support documents.

I'll get back to this thread in a while.

Anyone interested. PM.

[Iggy]

Iggy
Is it an "opinion" that you can hold a balloon and fly to mars?

Have you ever helped design anything?
I did. 68030 based motherboards.
This is not rocket science, it dated, relatively primitive electronics.

Comment all you want Magnetic. I'm already pulling in info.
604e not 603e, right?

[Iggy]

Iggy

Ok buddy, you should start taking pre orders too

No, my company actually produced the hardware we offered and we never took advance orders.
Besides, I didn't say I wanted to produce these.

I just want to find out how they were designed.
Difficult, not impossible.

Problem one, though. No new 604e processors. NOS yes. But only the 603e remains in production. Even e300 cored products like the 5200B (used in the Efika) are labeled as "not recommended for new designs".

[Iggy]

While I appreciate the "impossible challenge" just as much as the next person let me understand the reasoning. The end result in such a board is what, to run OS4 correct? If so we already have expensive options for that available now. The only other reason I can think of is to run 3.x PPC software. I didn't try any of these apps when I had my CS-PPC and Blizzard-PPC. Is there some killer app worth all this time and effort that won't run on the currently available options?

I really am just curious, not just trying to start something.  It seems like all the work has already been done for some years now.

Thanks for the open question, Jeff. Additional scepticism I don't need right now.
Actually, the only point would be to expand original 3000 and 4000 Amigas.
From a practical viewpoint, this would not even require Phase5 compatibility.
The only advantage to PowerUp compatibility would be the ability to run AOS4.1 and MorphOS 1.45 immediately without new ports.

As a current MorphOS user (and someone with an interest in OS4) I can easily see your point. I've been encouraging legacy users to move to NG platforms for years.
This entire exercise is to intended to be practical. Paying over $1000 for a 200Mhz accelerator for your A3000/4000 isn't practical either.

So I looking at options that would slightly increase performance and at least halve the price. About a 400Mhz PPC and 256MB of RAM.

[Iggy]

604e in the Cyberstorm, 603e in the Blizzards.

For a new board:

FPGA for 68K (to boot the system and run OS 3.9)
PPC SoC like the CPU in the SAM 460
RAM on-board
SATA on-board
Video on-board (if you can fit it like Sam 460)
USB and LAN on-board

Basically a computer on a cpu card.

Then i would buy it ;-)

Thanks for the input HammerD.
I like the idea of an SoC too. I already dug out the specs on the 5200B and will look at more capable chips (the only advantage to the 5200B is that its e300 core is 603e compatible).
I'll talk to a few people about the FPGA. This may be the right direction to pursue (but obtaining a '060 compatible core would be the main problem).

Two parallel avenues to explore now. Phase5 PowerUp compatible (but enhanced) and totally new. Let see what makes sense.

BTW - Looking at the size of the processor card your ideas seem possible. Video might have to be an add on (but then it was on the Cyberstorms as well). With an SoC everything else should fit.

[Iggy]

Thanks again for the input today guys. Both from forum members here and elsewhere.
After examining what documentation I could find I've decided that Magnetic (amongst others) has made some good points.

So I'm NOT going to examine any Phase5 hardware to review the design of the CyberstormPPC. Thomas Dellert can keep his outdated designs.But that doesn't mean I have conceded to defeat in this.

More to follow in the next few days. I promise to sum this up better by Friday.

Goodnight
Jim

[Iggy]

Iggy glad you finally saw the light I would hate for you to take on an impossible mission...

Well, I still hold that it not impossible. BUT, I often posted here that while you CAN do something SHOULD you?
After I got some information of the amount of programmable logic on the CSPPC, the amount of work needed just looked too massive.
The card needs a re-design. Much of that logic is obsolete and you would probably want to combine it all into one FPGA.
Heck, thanks to the Virtex 5 series you could even incorporate the PPC  into the same FPGA (maybe even the 68K core as well).

I'm not adverse to good advice, so I considered what you'd said, Jeff's comments, and a really nasty post from Thomas Dellert that Andreas dug up for me

Using Phase5's original work doesn't make much sense.

many many times have semi serious ppl in the amiga community  contempleted a new ppc card. Nobody has done it as its not worth the  time and effort by a longshot.

GOOD point. I'm into pain, but THAT much?

btw why do you have this commdrscot in your sig?

For awhile I was convinced Franko just wanted to be a contrary sod, but he has made important points about free speech, the necessity of NOT holding your tongue, and the HIGHLY questionable claim that ANYONE owns the rights to Workbench/AOS.

So William Wallace was right, freedom is most important.

@Karlos - You know what the real pisser is? Thomas Dellert not only would rather burn the Phase5 documentation then release it, but he probably has a G4 design that could be updated more readily then the 604e/603e designs.

I've got every bit as bad an opinion of him as he does of the Amiga community (worse actually). HE blames US for HIS business failures.

[Iggy]

As far as I know, the 603e is a BGA package, so with the right equipment, taking it out and putting a faster one in is fairly straightforward (without the right equipment, don't even attempt it). After that, you basically need to change the multipliers and clock crystals accordingly. IIRC, the multipliers are set via a block of four SMD resistors that you'd need to rearrange.

As tempting as it is, I've resisted the urge to fiddle with my BlizzardPPC. They aren't cheap to replace.

Yes, BGAs can be replaced and there are companies willing to do the work at such a low price that even if I had the equipment I would probably farm out the work.
Statchu100 has played with CSPPC upgrades, but it hardly seems worth it. 333 Mhz for the PPC? You'd hardly notice the difference.

I didn't think a G3 would work as a drop in.

Freescale's 5200B has a compatible (with the 603e) core , but would not work as an upgrade to an existing CSPPC.

[Iggy]

It's a 43% clockspeed increase over the fastest CSPPC ever sold by phase5 (233MHz) and a 66% increase over the much more common 200MHz version. Hardly insignificant.

I think its relatively insignificant. I remember similar speed differences in the x86 market. You'd never get me to go through that much trouble for those kinds of gains.

Unless you were just doing it for the shits and giggles of it. THEN, it makes sense.

[Iggy]

It's a 43% clockspeed increase over the fastest CSPPC ever sold by phase5 (233MHz) and a 66% increase over the much more common 200MHz version. Hardly insignificant.

Actually, my other statement was a bit rash.

With the X86 platform I didn't get that much performance boost with parts from above 200Mhz to 400Mhz.
But I bought a 450Mhz K6-III and was quite surprised. I spent a lot of time installing/recommending the cache on processor K6s (K6-III, K6-III+, and K6-2+) to friends and business associates.
The re-assignment of the 2X multiplier to 6X was great for both overclocking and for use in older machines that didn't have a 100Mhz FSB or higher multipliers.

As many came back to me, I had a lot of K6+ sales on Ebay. Almost all shipped to Germany (obviously the land of fanatics who must see how far things can be pushed). Also understandable as AMD had a Fab at Dresden (now owned by Global Foundries).
 
Due to my obsession with the last Socket7 processors , I was late in moving to later CPU architectures. However, when I bought my first SocketA processor (a low end 600Mhz Duron) I stunned.
At 600Mhz the Duron way outperformed the K6-III/2+ (and it could overclock to as high as 800Mhz).
So the change in design outweighed the pursuit of higher clock speeds.

It was a useful lesson when the P4 was introduced. I dismissed that line from its introduction and eventually gain my vindication.

With a successor to the CSPPC, we need something like what you were looking forward to (the G4) or at least say a 550Mhz 440 core.

And any compatibility will have to be software only. I don't need Thomas Dillert's sore feelings dragged into this.

Take care Karlos.

Jim