He does this every time there is a discussion with new hardware. He thinks he can do it better and read PDFs and look at public website price lists and everything Genesi ventures to produce will be too expensive, too slow, and not up to his standards. He plays the numbers game that many die-hard PC gamers love to play; the ones that actually bought two Voodoo 3 SLI cards for the 1% frame rate improvement. The one where 1.3GHz is not as good as 2.8GHz Quad Core. Where if it is not 64-bit it is irrelevant. I'll note that even Windows Vista still does not come in a cheap 64-bit edition, and most Mac laptops use 32-bit Core Duo chips, but he doesn't care about average, and usable, but "the best so he can brag to his friends".

It's the kind of thing I hoped to have left behind from the Amiga community, where the ABILITY to display 4096 colours was better than VGA at 256, even though you could not display that many in a game and still have it animate at a decent frame rate, and most Amiga games ran at 320x200 and not the 640x480 of common PC games, and also did not have CD quality soundtracks or voice actors.. :D

Given that he cannot produce a really good technical, financial or even a business argument for the G3, northbridge, seperate graphics, and resolve the issues involved with sourcing all the components... I tend to discount everything he says, these days.



Let me make this very, very clear.

NONE of you have ANY idea what the price of an MPC8610 actually is. Any estimate you find or price you see on Freescale's website - if that is even correct, did you not notice that all speed grades cost the same? - is certainly nothing to do with the price we *negotiate* with Freescale.

Please stop debating how "cheap" the MPC8610 may not be based on your uninformed opinions. You don't have access to all the information, and I'll delete any further posts that propose we use a $20 G3 or state worries about the totally imaginary "extremely high cost" of the SoC. I'd like to keep the discussions here at a productive level and not just nitpicking about whether Genesi is "doing with the right chip" or not.

That's good news!



OK, will do.



I understand how disturbing these things are to you, but deleting posts is an extremely unpleasant habit, that will make people flee away. We are very few people here, so loosing members is very bad.



I wasn't nitpicking, although I understand that it can be interpreted as such. I also recognize my posts are quite useless in the end.

So I'll let you do your job, instead of wasting your time and making you angry with my ramblings. Again, we are very few people here, and if we disturb the two or three guys that actually make things...

A final stab: Make a $200 Efika 8610!

(me ducks) 8-D

Sometimes it's necessary. I even delete my own posts from time to time :D



That wasn't directed at you.

There is a difference between being a critic and being critical; Arno I think, lacks the judgement skills he really needs to make an evaluation of a product. He also lacks the products - I don't even have an 8610 board, so how can Arno tell us it's too expensive and that a $20 IBM processor would be better?

I think the whole discussion of what chips to use is rather moot, considering we already picked these designs and are producing them. What is the benefit of going down the path of a long-winded debate on the price of the MPC8610?

Freescale don't list prices on their website anymore. When they did, all 4 speed grades listed had the same price. Is this something you can make an accurate judgement on how expensive the board might be? Is it really productive to discuss how cheap a system might be if it used some other chip?

I am not exactly driving a tank through Tiananmen Square here, but I am putting my foot down.



That would be a fine goal, something I agree we should work towards; after all, how could an Efika 8610 board sell at much higher if we had competition like the VIA EPIA CN (~$180)?

Hi Folks, be sure we are working closely with the Freescale e600 Product Manager and his Director to develop the best plan possible. Naturally, this includes much better pricing in volume than the numbers you have in consideration here. Also, there is more to think about: we still need a southbridge. We need to find the best one, which means the "best value" so we can develop the most efficient and low cost solution possible.

Details to follow...

R&B :)

The ULI1575 looks great on paper but I have heard alot of swearing over them from people who actually have boards with them. It's also hard to source from nvidia, since it's not really a product they seem interested in selling in volume.

I'm not aware of many PCI-e alternatives. On the other hand, with a SoC there isn't a whole lot needed from the southbridge, discrete components behind a PCI bridge might in some cases be a decent alternative.

The PA Semi Chitra eval boards use descrete SATA/USB components, the main feature lacking beyond that for a desktop system is sound. USB-based sound adapters work but they're not great.

I agree, audio by usb for default is not the best idea. Many usb
stacks have problems to sustain high isochronous transfer demands.
On the other hand I guess usb is a good upgrade path for many devices
(incl. audio) - but it depends on a good implementation. Thus, I'd
consider usb audio as option, but not for default.

AMD have several - SB600 and SB700S/SB710 are pretty much everything needed except ethernet. While not as integrated as some other chipsets (such as the ULi or SiS ones which have ethernet and all kinds of legacy I/O), I think the problem of sourcing a working PCI ethernet controller is not ACTUALLY that big a problem.

There are plenty of others around though, some embedded solutions, some more PC-market, some not actually PCI Express (the MPC8610 has a working PCI bus after all as well as two PCI Express controllers).

The one that that is for sure is using discrete chipsets for ethernet, disk, usb and adding audio on top, starts to negate the benefits of having an SoC in the first place.

>He thinks he can do it better and read PDFs and look at public website price lists and everything
>Genesi ventures to produce will be too expensive, too slow, and not up to his standards.

You are being paranoid and wrong.
My opinion on that matter can be found here.

>He plays the numbers game that many die-hard PC gamers love to play.

Nah, thats hammer. ;-)

>"the best so he can brag to his friends".

How do you even know I have friends? :^)
Before it gets ridiculous, better stick to topic and discuss actual matters not characters.

Yes, please. We need you both involved here. Thanks.

R&B :)

great to hear there is a new efika in the works. i dont have any efika now because my ibook has more power than that. i was looking at getting something faster. not sure how the mpc8610 compares to the ibook, it would definitely be faster than my current desktop which is a pentium3 at 799 Mhz. i have been searching for a new desktop for a while now to replace it but dont want x86 DOS junk. i have been reading up on powerpc systems and was hoping that the next efika would use a 64 bit dual core 2Ghz from http://www.pasemi.com/processors/pwr.html. perhaps cost is an issue. was there any thought on using that for the next efika ?

i have been re reading the post here and hope is was ok to bring up the pasemi cpu. sorry if it wasnt. honestly, i am looking for a modern computer that is not infected by DOS, intel, or microsoft. and now that apple has gone with intel, i am having a hard time finding a nice new ppc :(

matt

Maybe IBM has something up their sleeves: http://www.powerdeveloper.org/forums/vi ... php?t=1386

Well, before we get into this, I think it's best to say that.. the Efika is the new Pegasos, or at least the brand we wanted to make the Pegasos into, with a little refocusing..

So, I hope there will be many more Efika models in the works, so much so that it gets as boring as Apple increasing the processor speed on their laptops year on year, and there is enough choice you can just say "I will get an Efika to do the job" no matter what job it is :D



What model iBook is it?

What you have to remember is that even at 1GHz, the MPC8610 will be faster than any processor Apple ever put into any of their Power laptops. Apple never used a bus speed faster than 200MHz, and they never had a chip manufactured at 90nm. They stopped at the 7447A and bumped core speeds (up to 1.6GHz, using some fairly.. let's say unethical.. bin sorting techniques)

It should not be underestimated what almost tripling the memory bus speed can do for the G4, especially for AltiVec which has traditionally been a unit you could never feed with enough data.



Well, PASemi would be a great thing to use, but we need a good reason to turn a very dedicated networking processor into a refined design like the Efika..

Matt, what's about the price, MPC8641D compared to MPC8610?

And a second question, what's the fastest chip FreeScale can ship? MPC8610 (1.33GHz?) and MPC8641D (1.5GHz?)

Thrid question :D
How does a 8610 1GHz perform compared to a 7447 1.5GHz (estimate)?

Fourth question :D
Whats the difference between 8641 and 8610? Besides Cores/MHz and so on ;-)

Thanks
Karl

I don't have figures I can give you. The MPC8641D is obviously more expensive.



Yes.



It depends on the application workload, but the MPC8610 will trounce the 7447 in power consumption, and anything that requires a lot of memory bandwidth.
With AltiVec, that was always important, so you'll obviously see the benefit. I'm itching to get Konstantinos to run his freevec benchmarks on one.



Built-in Ethernet ports in the MPC8641D, and the DIU in the MPC8610. There are some obvious improvements to the power consumption, and any updates to the IP cores for PCI Express and so on that come with having a couple years of development between chips. That's about it.

Right now, we think because of the complete lack of dual-core benefitting software for Power Architecture, the MPC8610 with it's far lower power consumption and integrated graphics (none of thedoubling or tripling the system power profile just to get a display we get with the Efika or even Pegasos) it is a far more interesting chip.

I personally want to see something around the MPC8641D later. The MPC8610 makes a safe, very interesting reference though, keep it simple and effective.. once the benefits of the chip are evident, the further benefits of the MPC8641D can be exploited more fully.

Will the 8610 board have standard PCI/PCI-e slots and will it be able to fit in a standard case (position of screw holes, backplate, etc)?