We are about a week away from finalizing our Best of Show nominees for the Freescale Technology Forum. If you would like your development to be considered for submission, please write us.

FTF is 16-19 June in Orlando, Florida. If we select your Project for submission we will bring you to FTF as our Guest. Submissions can include software that runs on Genesi hardware (Pegasos/ODW, EFIKA/Open Client) or Freescale hardware we/Freescale have sponsored (5121e ADS, 8610 or ColdFire Development Systems). Submissions can also include a complete system package that you have developed.

Please let us know!

R&B :)

_________________
http://bbrv.blogspot.com

Thanks for all the ideas and inquires, but we have decided not to participate in the Orlando FTF event. The main reason is that the next version of the 5121e and first version 5123 won't be ready before the Show. Consequently, it will be impossible for us to have the new designs ready with the latest release of the silicon.

We will be participating in the Global Events which begin in the Fall. We will be writing those of you that have written us as that time gets closer. Thanks again for all the ideas and support.

Best regards,
R&B :)

_________________
http://bbrv.blogspot.com

Big, big miss.

The 5121E is not ready? What happened to those working LimePC prototypes from Chinese THTF? We all saw them.
I'm very surprised that a new version of a chip is needed, after seeing working prototypes almost looking like final product. Which are those much needed differences? Was the original version so bad?

Will the new version of the 5121e have cache coherency, or is that too much of a big design change to be included in a mere revision of the same chip?

We had the same issues with the MPC5200B which delayed that Efika by 6 months and more. Just because there is silicon and it runs code doesn't mean it *works*. We had serious bugs in the original MPC5200 which delayed until the MPC5200B, and then in the MP5200B there were some serious BestComm and ATA issues which you'd have to resolve in software (hence the DMA-PIO driver, but this took a long while to find out).

The MPC5121E is not what we wanted to start with but it's a working, usable chip nonetheless. However it is not ready for users.

THTF's prototypes at CES were just that - prototypes. A working display model running SOME code but not necessarily doing anything exciting. There is absolutely no chance you would be putting that device into production with the current state of the silicon.

_________________
Matt Sealey

As always, thankyou very much for the information, Matt. Lesson learnt!
When Genesi does something, you can be absolutely sure that it is not motivated by any kind of bullshit: A technical compromise exists.
From what I've learnt recently, cache coherency is not a big issue itself. Many CPUs don't have it. You can write programs anyway, in this case, kernels.
It merely complicates things if you want to make a kernel that runs on a variety of CPUs, and some have cache coherency while others don't. You would actually need different kernels.

The answer is no, the 5121E (and 5123) are not designed with cache coherency in mind. It is, as you said, a software problem for the most part. The software simply has to make special effort to get the best performance. Of course in certain scenarios cache coherency can make systems slower - snoops and acks and cache management take up valuable system resources. Usually chips built this way don't have too many peripherals though.

We took exception with the 5121E because it has a LOT of peripherals and many of them are driven by a high-speed DMA controller. This will make development a little more hassle than it could have been (and than anyone is used to on PowerPC) but, it is all in a day's work.

At the end of the day, the userspace programs - anything you install outside the kernel, will be exactly the same, so it doesn't affect you guys much.

_________________
Matt Sealey

I just saw this patch the other day - does this mean the problems are finally solved?

...by now you have seen the FTF Keynote.

;-)

Don't worry. We have not 'missed' a thing.

R&B :)

_________________
http://bbrv.blogspot.com

No. This didn't fix any problem at all.

_________________
Matt Sealey