@ Raf_MegaByte

Quote:

No it hasn't.

And I don't think it will be released either (at least I hope not). There is a huge difference between talk on a forum, and actual consumer-level production runs, building up a support infrastructure, etc.

First of all, the Articia S is a *really* old chip now. It was designed mainly with the old class G3's (and G4's? (not sure of that)) in mind that was available back at the time. The Modern PPC's we have today was science fiction back in the days when the Articia was designed. Even *if* the Articia would have worked as planned (and advertised), I don't think there would have been any guarantees it would work fine with the 7448. New generations of PPC CPU's are generally acompanied with new generations of system controllers. This is logical for a performance point of view, but also an answer to differencies in low-level characteristics in the CPU's as evolution moves on AFAIK. In the Freescale HPC-1 platform (evaluation board for the MPC7447A CPU), a Marvell Discovery III was used. In the Freescale HPC-2 platform (evaluation board for the MPC7448), a Tundra TSI108 was used.

But the Articia S *didn't* work as planned (and advertised), not even back then. In fact, it ended up *notorious* in the industry for its flaws and limitations. When Genesi used the Articia S in the first Pegasos design, they had planned to release the Pegasos at a certain date (can't remember when exactly), and they originally planned for a 1GHz G4 CPU option. In the end they had to postpone the release at least half a year, they had to sell it as "betatester" systems under certain conditions, and only with 600MHz old class G3's (a 1GHz G4 module was simply out of the question). Later they had to release two generations of hardware patches and expensive trade-in programs, all thanks to the Articia. Among the last things they did before throwing the Articia in the trashcan, was to try the very latest Articia revision coupled with the April2 fix, and a modern class G4 (I think it was the 7447). They found out the CPU's/Articias only worked in certain random(!) combinations (which made it impossible to sell the G4 as a general upgrade card, it had to be sold in tried and tested combinations of motherboards+CPU bundles), and not *entirely* satisfactory. That was a 1GHz 7447. Even Eyetech downclocked their old class 933MHz G4 CPU's to 800MHz (which is a very odd thing to do if you think about it, when there was an option to increase the price with 16% accordingly, or simply say "hey, here you have more value for the same money"). There has been an innuendo (tons and tons) of reports of strange A1 behaviours, data losses, and other artefacts of the Articia (also in a random manner, it seems). I think it was "Slash" over at AW that tried many times to compile an entire Linux distribution on his A1 (which demands *a lot* more of system integrity and stability than casually running Ibrowse an hour a day or similar). This takes a lot of time with full throttle, and he could never get through it; there were always a crash somewhere along the way. When he *underclocked* his CPU to 600MHz, the situation improved greatly. There are reports of strange behaviour from the modern class G3 (750GX) CPU's as well, also in a random manner as it seems.

I think there is a clear difference in using old class CPU's at low frequencies on the Articia, than modern class (like 7447 and 750GX, not to mention 7448) above 600MHz.

And now we have this new bold announcement from "Adam of ACK", a guy who announced the "Power Vixxen" in December (that is December 2004, *not* 2005) and still hasn't been able to show a working prototype, heck not even a simple photo, or *any* kind of sign for that matter that suggests that this is nothing but vapour. And here he comes again, with a new product (before even showing signs of life from his first one). This time he has *reverse-engineered* a Teron 7455 CPU card from the brilliant brains of MAI (or whoever made the "brilliant" work with the Teron HW designs (spot the irony)), and wants to slap on a *1.7GHz 7448* to the Articia S! I mean, wow!

I'll tell you what; if you put your ears close to the motherboard, and the room is silent enough, you will probably be able to hear the Articia brainfarting through the ethernet connector in pure horror!

It would also be interesting to see what kind of warranty ACK will give its customers. If peoples $850 investment doesn't work as supposed on their A1's, will they get a full refund? What happens if their CPU cards breaks down in a year and a half, will Adam still be here by then, offering full repair/replacements, free of charge? What kind of financial backup does he have, in order to secure a long term commitment to support and customer service? I'm a bit surprised that no-one is asking these kind of questions.

Seeing how he is requesting e-mails before production run is not a good sign in this context IMHO. I see a new "Eyetech situation" coming up here.

BOHICA!

Bend Over, Here It Comes Again! And everyone jumping of joy among dancing bananas about it.


Quote:

The price isn't only "extremely high", it's outright ridiculous!

For that kind of money, you can get a complete ODW, with

* Pegasos II motherboard *including* a 1GHz G4 processor
* 512MB DDR RAM
* 80GB ATA100 Hard Disk
* Dual-Layer DVD±RW Drive
* USB 2.0 Expansion
* ATI Radeon 9250 graphics - DVI, VGA and S-Video out
* Low Profile Small Footprint Case - Tower or Desktop Orientation (92x310x400mm)

and you would still have $50 left for pizza and beer!

If you want to compare apples with oranges, you can also have a look at how far $850 would take you in the mainstream x86 platform.

I mean, the $850 price tag is *insane*!

And besides not functioning properly, the Articia S is also crippled with bottle-necks all over, so this kind of CPU will be a complete *waste of money*, since it won't be able to even *come close* to its fullest potential (for which you pay your money).

Quote:

Nope.

The "market" for this product is so tiny that it hardly can be called a market at all, and moreover, it's fixed in size and will never grow any bigger (the A1 platform is stone dead and no more A1's will ever see the light of day). How many units can he expect to sell? 50? 100? Nah, not that many. Anyway, it will never be *anywhere near* "a certain amount" (as you put it) to get attractive volume pricing, so no, the price won't come down in time (unless he is forced to sell out unsold stock at great discount at some future point in time).

The only way to make the price come down would be to establish good relations with the CPU manufacturers (á la Genesi) instead of ordering some handful of units at list prices from some online retailer, also being able to manufacture the thing at low cost in a "in-house manner" (á la Genesi), but most of all to have a realistic views/expectations of your financial returns from a thing like this.

I'm sorry for my harsh words here, but this price is a bloody rip-off!

Quote:

But those are totally different cards! Besides, this wouldn't change the volume pricing aspect at all, since the Pegasos is no high volume platform either ...

Quote:

They don't want to build 7448 cards?

Quote:

If he really wants to, Adam can manufacture CPU cards for the Pegasos *without* Genesi's permission. He could get a flying start with the Pegasos2 design that was released freely on power.org. And when the upcoming public firmware update is released, chances are that it will *work* as well ().

But I'll tell you what - if he is planning to actually go ahead with this $850 price tag, then he should stick to "teh REEL Amigans!!1!1!" and leave the rest of us alone. I think this suits them better ...

Quote:

You must surely be joking now, right?

By now, Genesi have low-level experience from both the 7447A and 7448 (the Freescale HPC-1 and HPC-2 evaluation platforms) since some time, that is quite unique in this context (in comparison to some other people) IMHO. They have adapted their firmware for these CPU's (and northbridges) a long time ago already (they even sell 1.4GHz ODW's based on the 7447A). B-plans expertise in HW design and their high standards is well documented and acknowledged. They have had production-level 7448 CPU's in-house since early this year; the 1GHz version is used in the High Density Blades. Do you really think Genesi would need to rely on "Adam" (a guy who *actually believes* he can slap on a 1.7GHz 7448 on a reverse-engineered 7455 CPU card and make it work with the Articia) in order to get some 7448 modules out for the Pegasos? Of course not, that thought is ridiculous IMHO.

No-one knows the Pegasos the way Carda&Knäbel does (no-one knows the Articia S as good as them either BTW, please think a little about that ), no-one has the relationship with all key players as Genesi has (component supplies, manufacturing, etc), and no-one besides Genesi has yet shown any ability in creating and (more important) *supporting* end-user hardware in a proper way.

Relating themselves with Adams little experiments on living humans (á la Eyetech) can only result in their name beeing dragged in the dust when the project crash and burn and the consumers are left stranded.