community update, Thursday, January 10, 2008

[Michael Shiloh]

Hello,

A brief status report from OpenMoko:

Consumer Electronics Show, Las Vegas
------------------------------------

We've just returned from CES where we showed the Neo FreeRunner (GTA02). 
Although this was still a prototype it performed fairly well. The UI we 
had installed was the same as the most recent for GTA01. Of course most 
of the press was interested in this as a consumer device. Nonetheless, 
interest was very high. A number of Linux and Open Source enthusiasts 
came by, and of course they were thrilled. Most of them already knew 
about this project but wanted to see the GTA02 and to hold it in their 
hands. We were also visited by some Linux luminaries (Doc Searls, 
Maddog), which is always very thrilling.


GTA02 progress
--------------

We're still testing the hardware and gathering up little issues before 
determining whether we need to create another version of the board. We 
still expect to start shipping Neo FreeRunner sometime in the next few 
months. As always, we can't be more specific, because we're not sure.


Neo FreeRunner press release and the CPU speed
----------------------------------------------

The Neo FreeRunner press release that went out last week indicated a 
500MHz CPU which, as many of you pointed out, is in conflict with the 
400MHz stated on the wiki. I researched this among the experts and I 
think I have gotten to the bottom of it:

Background: The GTA02 hardware was designed for a 400MHz processor. The 
particular combination of CPU speed, FLASH size and RAM size that we 
wanted became unavailable, so a 500MHz device was substituted.

Although it would be nice to take advantage of this faster processor, 3 
problems make this impractical:

1) The GTA02 hardware was designed for 400MHz, and running at the higher 
speed may uncover any manner of problems, e.g. timing or drawing too 
much current. It might be fine, but needs to be tested and analyzed 
thoroughly.

2) At a CPU clock of 500MHz, we have to use a slower memory bus clock 
than we would if the CPU were clocked at 400MHz. The memory bus is 
clocked at a fraction of the CPU clock. A limited number of ratios are 
selectable. At 400 MHz, we could use 400/4 = 100MHz, but 500/4 = 125MHz 
is too fast for the memory, so we have to go to the next larger divider, 
which is 500/6 = 83MHz.

3) Power consumption does not rise linearly with CPU speed, and at 
500MHz is extremely high, resulting in extremely short battery life.


That's all for now. As always, I welcome your feedback, questions, 
comments, and concerns.

Michael