(Backstory for those unfamiliar with the Raspberry Pi board: The
Raspberry Pi Foundation was formed by a group of industry leaders in
Cambridge UK who believe that decreasing access to easily-programmable
hardware is contributing the decline of interest in CS/IT among young
people - compare the number of machines that booted to a programming
prompt in the 80's with the number that do today. Their solution is to
develop and sell a computer that's cheap enough and open enough that
youth, schools, hackers, makers, and so forth will use it widely. Their
first device is a $25/$35 ARM11-based computer).
I've been working with a Raspberry Pi foundation
) and would like to see us (Fedora ARM) provide
first-class support for their device. The foundation kindly supplied an
alpha board, and we (at Seneca) have got F13 running reasonably well.
There are still some significant hurdles -- for example, it has awesome
1080p high-def streaming video and OpenGL performance, but X11 is still
painfully running on fbdev -- but these can be overcome, and I think
that this will be a useful and important device.
The Pi has a really strong GPU and an ARM11 (armv6) processor on a
BCM2835 SOC. I anticipate that this device may sell into the millions of
units in the next two years, and having Fedora as one of the primary
operating systems will be beneficial both to the Pi users and to the
The firmware blob on the GPU side is closed-source, but that side of the
chip can be effectively treated as hardware; lobbing stuff (H.264,
OpenGL, ...) over to the GPU through a socket-style interface causes
stuff to happen. The kernel interface to the GPU side is open source. On
the ARM side, there are a few small userland pieces that aren't yet open
source, but hopefully will become so; therefore we'll initially need to
provide a Remix (rather than a Spin) for the Pi, and should aim to
switch this to a spin as soon as possible.
The expected ship date for the Raspberry Pi is (early?) December
(initial 10k units).
If this device ships in quantity -- and it looks like it will -- I think
we should at some point look at providing optimized armv6tel+vfp2 builds
of key packages to improve performance; a student here is checking to
see what level of performance benefits this could provide.
My current focus is on trying to adapting the glamor X server to work
with the GPU, while a few other folk here work on optimizing other
pieces (boot, browser footprint, and so forth). We'll keep the list
apprised as this progresses. Jon Masters and I are working on getting
Red Hat and Seneca to fund the purchase of a few dozen boards from the
first production run to distribute within the Fedora community.
(These are exciting times for ARM! - Raspberry Pi and BeagleBone
announcements on the low end, Calxeda and armv8 announcements on the
high end :-)