Hi, the 8640D is probably very closely related to the 8641D. The max power numbers may have been reduced and there could have been frequency changes, but the two chips are likely to have the same pedigree. In our opinion, this is a story about marketing existing technology that has improved progressively with revision over time. It sounds like a great thing and it probably is.

R&B :)

P.S. The 8610 development system finally shipped to Konstantinos last Friday. We will still be using the 8610 for the netbook we are developing.

_________________
http://bbrv.blogspot.com

And the 8610 is a great chip. Those, who had seen the 8610 reference system in action at SceneCon 2008, wanted to get one ASAP :-) It is fast, while also energy efficient.

_________________
CzP
http://czanik.blogs.balabit.com/

Yeah, like the "G4" -> "e600". New attention to already existing but over time "invisibly" improved technology. New name to stomp off a new start.
I'm sure it is! :-) While it lacks the 8610's display controller, it definitely has a lot of advantages in other areas.
That's good to hear! Looking forward to see more info about that netbook project of yours, and the 8610 seems like a perfect match for a powerful mobile application! :-)

And then at some point *maybe* we can have a successor to the Pegasos mainboard. Before all remaining pegs breaks down, disappear and becomes forgotten on attics or in garages, or becomes totally obsolete regarding peripheral connectivity standards. Every developer needs a desktop!

;-)

...and it seems we now have a Freescale CEO that understands this, which is of course a *very great* thing for Genesi and Power Developers.

We have not been so motivated and so busy for at least a couple of years.

R&B :-)

_________________
http://bbrv.blogspot.com

Are there any photos, or better, videos about your presentation?
I agree that the MPC 8610 is a very, very clever chip. There were early talks about supposed incredible abilities of its integrated display unit ("DIU"), but now Genesi will pair it with a Radeon chip. Interesting...

Well, good question :-) The presentation had a live broadcast, and many pictures were taken. But I could not find (yet) any of these on-line at http://scenecon.org/ AFAIK, the video will be uploaded with English subtitles, just as last year. I have a strong feeling, that I forgot to talk about a couple of topics, as was talking about the displayed boards almost all day long and had wrong assumptions during the presentation, what was previously mentioned. We will see, when the video is on-line :-)

BTW: DIU is really great. With a completely unaccelerated display driver, I could still play 'Big buck bunny' in 720x1280. But it has some obvious shortcomings, like no 3D at all.

_________________
CzP
http://czanik.blogs.balabit.com/

Exactly nobody said exactly that.

Depending on the performance requirement and feature requirement of the product, the DIU might not be up to scratch; it certainly does not do 3D and has a maximum resolution which is not too huge, plus the more data you need to push through it, the more system bus bandwidth is taken up.

Sometimes you will just need a discrete graphics controller. I personally think the DIU is good enough for a lot of uses, so in the first design, perhaps it does not use a discrete graphics chip.

However when you start moving to higher clock speeds, higher screen resolutions and needing more advanced features where a real GPU would be required, then this is where we'll go, and traditionally we've focussed on Radeon, but it could just as well be an nVidia chip, or something you never ever heard of.

_________________
Matt Sealey

Maybe not, but at least it was implied here. Or maybe they meant to say southbridge?

Anyway, looking forward to hear more about this in the formal announcement! :-)

I'm confident that we won't need a "full" southbridge for a Netbook design.

The graphics thing is still up in the air. It depends how much battery life you want out of it. A mobile graphics solution can take as much power as the CPU itself. That instantly, actually, halves the battery life. Perhaps there will be two models, perhaps not. How I'd put it is "[IF we use discrete graphics] we will use ATI Radeon". It just so happens they have the best mobile chips (nVidia run really hot) and potentially the best free software driver support (and potentially also the best proprietary driver support)

Power management and battery life is very much in discussion right now, and whether we can support graphics, which chip we'd choose (this is tied to current driver support and future driver support) and how this impacts the functionality of the laptop is very important. A NetBook with only 90 minutes of battery life would not go down very well. One with 6 hours, even if it did not have 3D...

_________________
Matt Sealey

Personally, I'd buy a laptop with 6h+ battery life instantly even without 3D. We could always optimize the hell out of the 8610 DIU anyway using Altivec, even 3D would be more respectable :)

Same here! 6+ hours would be absolutely fantastic, even more so on a system with an Altivec unit to play with :-)

I see myself using it like having an embedded system/crypto development lab in my pocket while traveling the 5 hours to Helsinki for a conference or something.


Johan.

_________________
Johan Dams, Genesi USA Inc.
Director, Software Engineering

Yep, I have a blog... PurpleAlienPlanet

Good to hear! Let's see if/how this will pay out in future CPU development as well as support down the line.

Well, the 8640 is now available, and I just noticed they sell the "HPCN" evaluation board re-fitted with the 8640 at a "promotional price" of $1499.00 USD (valid through September 30, 2008). While still being an evaluation board, it's definitely provides what a developer would ask from a desktop computer. A slot for a PCI-E x16 graphic card, USB2, SATA, Gigabit Ethernet, Sound, etc.
That's great to hear! :-)

100% agree! But would it be profitable? Are these optimizations easy to immplement? I think high end 3D performance is not required in this kind of computer.

easy? well they're not trivial, but they're not rocket science either. Considering I've already done most the 3D stuff (check http://www.freevec.org/category/simd/al ... operations for more) it is a case of integrating these functions in Mesa -which is something I've already started anyway. But right now I'm more involved in rewriting optimized functions for common trig functions (sin/cos/tan/etc and functions like exp,exp2,log,etc). The benefits are huge even without Altivec (tan is 2x as fast with full IEEE754 accuracy, same as glibc, sin is 50% faster, etc). Patience. :)

Using AMD's southbridge is a neat idea. I remember drooling over Uli's M1575, but Nvidia don't seem to like to talk about that anymore and it seems listed under legacy products there. While the M1575 proved that a southbridge to AMD's north can possibly be a standard PCI-Express endpoint device, whihc is desirable for these PowerPC SOC chips, that doesn't mean it's guaranteed that all these A-Link chips are standard PCI-Express endpoint devices. M1573 proved that...

Does anyone know for certain that any of AMD's own SB chips are standard PCi-Express endpoints? Or are they something close but not quite enough to work with anything other than AMD's northbridges?