We don't think so. Or at least, I don't think so.

Where you have problems is obviously where a new chip design hits and you think, that will be PERFECT for this idea we had! Of course this chip is still 2 or 3 months from engineering samples, and when you get it, you still have 6 months to do development WHILE finding errata that the chip vendor doesn't publish or even doesn't know about yet. Then, if the chip is not ready, you may sit on a design for 6 months, or you may sit for 6 months in conference calls debating how to work around a feature or if it is a bug that can or will be solved at all.

Some things, you simply cannot work around though, and an SoC makes it harder. If you do not have some chip feature embedded in the die or there is some quirk of a certain revision you cannot just do some clever trick to add it externally like the April fix.

We cannot just "fix" the UDMA mode on the MPC5200B ATA but software can be written to alleviate the CPU load problems of using PIO. This takes time, in a holistic view of system design which is VERY important to take into account. It is not "correct" to simply give up and say, we can add an external ATA chip, when this may cost more than the SoC itself!

The MPC5121E will be a challenge because it throws away certain parts of the MPC5200B - thankfully the ones that make you want to throw yourself out of a 4th-storey window - which means you are in line for a new set of problems. The chip is still a little way from release as per the dates published by Freescale, and design work is still being done on it.

The 8610, however, is a friendlier beast, with all the experience of the 3 years of development on the e600 core (7448) and integrating it into SoC platforms (8641D), most of the IP present in the chip being already present in a lot of other Freescale chips. It is not anything particularly new, it is just an exciting and potentially very cost effective new spin on things. It should be a no-brainer - I say that now, but we will see.

_________________
Matt Sealey

I wish Genesi and THTF best of luck in this project.

That's good to hear! I'm actually *thrilled* by the possibilities this chip may have to offer, if only the challanges can be overcome. This is edge technology! :-)
To me it looks like a "Pegasos2 on a chip"; a mix between a 7445 and 7447 class CPU cores judged by the L2 cache and top speed characteristics, but with improved memory performance and PCI-e. And of course included GFX, albeit without the 3D part unfortunately. This is nothing to frown at IMHO; I mean put this on a MiniITX or FlexATX (or MicroATX) motherboard and you would have a cheap(?) Pegasos class machine with upgraded performance end specs. "Every developer needs a desktop", right? ;-)

It's far away in time though...

It's just a nominally-clocked single 7448 with less cache. The fact that it is the latest version of the e600 core is important; it'll help AltiVec slightly.
Indeed. I am thinking why not just provide a system which has "integrated graphics" as with Intel or Via boards. It needn't be 3D and it needn't even be accelerated 2D (Linux console drivers are rarely accelerated, and X.org 2D drivers can be accelerated with AltiVec in userspace - in fact with libraries like Cairo and Composite, most of the acceleration requirement is alpha blending blitted blocks together, which AltiVec would be VERY good at (4 to 16 pixels per simple blending with saturation operation..).

The DIU controller can handle 3 fully blended planes (for example background, sprites and OSD) in 1024x768 and handle a single flat plane (no blending, would have to be CPU-blitting) at 1280x720. It's not going to please people with 19" monitors, but we will definitely be looking into providing for the users if a board comes of this.

AltiVec video decoding will shine on this. No special bandwidth required - just use AltiVec. Draw your subtitle plane, decode your video (and YUV to RGB conversion which is highly accelerable) to the correct resolution.. et voila. Doing simple windowing operations would be fine. Via chipsets have significant amounts of video decoding hardware to handle all this but AltiVec is more than capable especially with the right data format and with a 600MHz FSB at it's disposal.

Of course, if you want 3D you would go and buy a Radeon but I think we are getting out of the business of picking your graphics card. You can support yourselves just as well with a PCI Express slot and finding the right driver module..
Why conform with an arbitrary PC form factor standard? The 8610 is integrated enough that it MAY be possible to design an 8610 board with a PCI Express slot for external graphics or other peripherals, in the same form factor as the current Efika. You might say "but ITX is an industry standard", go and tell Shuttle that, and they will laugh in your face.

Of course then the onus is on Genesi to provide a case, and we have two designs already, one of which sold a few hundred units as the Open Client. A better, lighter case design with some pretty and shiny added on, shared between potentially 3 board designs..
This kind of board design would probably be done by the early summer and shipping from mass production by late summer if we started on Monday (we aren't, but as an example). Not that far away. The Efika has been out for nearly 14 months now.. it doesn't seem that long, does it?

_________________
Matt Sealey

Why make a non-standard form-factor one more time?!

If you do then I really hope you never sell any bare boards
to anyone besides large OEM-customers.

If you do what you did with the EFIKA you will have the same
nagging on forums and bad publicity about non-standard
form-factors. It would be very sad for Genesi. :(

Neko:
Shuttle is a multi-billion company. They never, sold their
barebone boards without any case. If you want their pro-
ducts youget the case with it. Period.

We didn't have much nagging. What they nagged about was that there were no cases for it.

If we design a case - or two maybe - then there is no reason to nag.

Remember NanoITX and PicoITX have literally THREE cases on the market which have the correct standoffs installed, and the correct cutouts for the backplane. Nobody "nags" about those.
That's not a good solution? The way I see it, you will either want an Efika with a case like a PC - in which event we will design cases for it - or want to put your Efika in some strange mounting some other way like in a kiosk or embedded environment, where the case will be bespoke.

Either way that means we just need a desktop-oriented case for it. I see the 8610 as potentially a Pegasos version of the Mac Mini or iMac - distinctly tailored for particular use. Some people may take out the Mac Mini board and put it in an ATX case, that is their choice.

However we should not make a 17cmx17cm ITX board and have lots of redundant space on the boards just to make a "standard" form factor for 3 existing case designs. We should not make a 12cmx12cm NanoITX board *simply* to conform to a design which literally has TWO existing case designs on the market.

There is little point going for a standard form factor unless the design calls for the amount of PCB space that roughly fits that standard form factor. You cannot find any boards outside of Via or very specific Embedded solutions which use ITX, and when you move out of the x86 market, ITX boards are exceedingly rare. When they do appear, they tend to be PPC systems (8349E ITX, MPC5121E ITX). If we make boards around these chips there is little point repurposing the same large form factor for a cost reduced, space reduced design.

Do you understand where I am coming from with this?

Remember; we NEVER stopped anyone creating cases for the Efika. Unfortunately just like the lack of driver work from users, the lack of hardware mods from users, the lack of work whatsoever despite the promises made in Project proposals, nothing gets done. This is not Genesi's fault. Do you want to make it better? Then design a case, and put it into production, and we will probably buy them!

_________________
Matt Sealey

We have before...

http://bbrv.blogspot.com/2007/03/case-study.html



R&B :-)

_________________
http://bbrv.blogspot.com

I understand perfectly what you mean. This doesn't mean that I
agree to what you say.

Sure, a non-standard form-factor for EFIKA 2/5121 will require
official EFIKA cases from Genesi. Unless you can deliver these
cases upon release of EFIKA 5121 there will be too many boards
just lying on tables, collecting dust.

I know the EFIKA (and probably EFIKA 5121) is "just" a developer's
tool for creating something bigger. But just because of this you
can't ignore complete solutions for said developers.

Look at Apple's early-bird Intel developer machines. Complete
machines in decent midi-tower boxes, ready for whatever you
throw at it. You will need to do the same for the EFIKA 5121.
These developers paid 3000 USD for this. The component
value was probably less than half of that, but they bought it
just because this was the thing they wanted. The extra dollars
was for the entire support around these machines and Mac OS X
with all possible SDK:s..

BBRV:
Yes, I know about the case. The case is good for developers.
The problem I'm referring to in this post is that it should have
been available from the start. And not some 6 months later.
I don't want you to do the same mistake twice, and therefore
please, do not sell the bare EFIKA 5121 in single-quantities
to non hardware/OEM developers.

Genesi:
All I want is what is best for you and the community. To me the
hardware and CPU is what is least important. What is most
important to me are solutions that simply work. I am ready
to pay a little extra to get the product I really want.

Offer me a complete solution that covers what I'm after at that
time, and I will buy it. I am sure that many here will do the exact
same thing. Heck, I may even gladly sell them in Sweden through
my company!

The product specs look great and cover a wide range of market demand.
Powerpc does need products/solutions in all these categories that are now dominated by geode/xscale/arm.

KUDOS to the Genesi people for their efforts to get things going, regardless of problems.

Stelios

Some time ago I found another interesting form factor: vesa mounted. I thought that was pretty cool 'out-of-the-box' thinking :-)

Via are certainly NOT the only people to have made a case that had VESA screw mounts on it.

They may well be the first to make so much fuss, however.

It's an interesting idea, at least, and it's something we can think about with a case design.

_________________
Matt Sealey

A movie with the limepc in action:
http://www.mefeedia.com/entry/limepc-li ... 8/5909024/

Unfortunately, that was not too good of a "demo" -- we will post some news about this effort in the next couple of weeks.

R&B :)

_________________
http://bbrv.blogspot.com

from GDA Technologies based on 5121e

http://www.gdatech.com/pdfs/G-MPC5121Ba ... atform.pdf

_________________
acrux _at_ linuxmail _dot_ org

Sexy! C'mon guys, not a single word about this? Has Genesi abandoned the MPC5121e project?