|
Latest Project Blogs
|
|
|
|
|
|
|
 Efika buildbot up and running!
Project: GNUnet Portability/Hardware Testing (Networking, Routing & Storage)
posted by Nathan Evans on 2nd September 2010
After some time getting the Efika MX set up for running GNUnet, we are proud to announce that it is now smoothly building GNUnet!
You can check this particular bot, as well as all of our current bots, at the GNUnet buildbot page.
Currently there are some issues with our testcases related to shared libraries (we believe this to be distribution related, not hardware related). Once we get all of our known good tests working we will gather and put up some benchmarking information at the blog here.
As an aside, the Efika is a great little box and has been relatively easy to get working.
|
|
|
|
|
|
|
|
|
|
|
|
Gnash on SmartTop
Project: Gnash optimizations for ARM (Multimedia)
posted by Rob Savoye on 1st September 2010
Now that I have Gnash building on an i.MX515 based SmartTop, I uploaded Debs for Ubuntu Maverick to the Gnash repository at: http://www.getgnash.org/packages/. You can grab these by adding this to your /etc/apt/sources.list: deb http://www.getgnash.org/debs/ubuntu maverick main
Just for grins, here's a screenshot of Gnash handling YouTube video on the SmartTop/ Next task is to work on the OpenVG/OpenGLES rendering backends to improve performance.
|
|
|
|
|
|
|
|
|
|
|
|
ARM assembler progress
Project: ARM Assembler for OpenJDK (Applications & Software)
posted by Rob Savoye on 27th August 2010
I've been making slow, but steady progress on fixing the ARM assembler used in OpenJDK. When the calling convention changed to fix another bug, it broke the ARM assembler, forcing Java to drop back to the ARM interpreter, which is much slower. The goal is to get this all fixed up in time for the release of Ubuntu Maverick.
Currently I've fixed all the low level assembler code, and it passes that part of the testsuite. Now I'm focusing on the C++ code that gets called for the thumb2 JIT. This should be easier than hacking the low level code, but it's still a bit of a register allocation nightmare. Most all the registers are being used, so finding temporary ones to support the fixed calling convention has been very interesting...
|
|
|
|
|
|
|
|
|
|
|
|
initial Gnash port
Project: Gnash optimizations for ARM (Multimedia)
posted by Rob Savoye on 27th August 2010
This isn't too exciting, but I've got the current Gnash 0.8.8 release working fine on the i.MX51. This version still uses software rendering till I make more progress on the OpenGLES/OpenVG support. But even with using software rendering, the video performance is adequate. So now i/.MX51 users can waste their time watching YouTube like everyone else. :-)
|
|
|
|
|
|
|
|
|
|
|
|
I'll present Efika MX Smarttop on Chaos Construction event
Project: Add support of Efika MX devices to kexecboot (Embedded)
posted by Yuri Bushmelev on 12th August 2010
I'm going to Chaos Construction this year with my hardware: Sharp Zaurus SL-C1000/SL-6000L, Ben NanoNote and Efika MX Smarttop.
Chaos Constructions 2010 demo/lan/hack party is going to take place on 28 and 29 of August in St.Petersburg, Russia.
Site of event (in Russian): http://party10.cc.org.ru/
Exhibition page: http://party10.cc.org.ru/exhibit.php
|
|
|
|
|
|
|
|
|
|
|
|
Efika MX arrived
Project: Add support of Efika MX devices to kexecboot (Embedded)
posted by Yuri Bushmelev on 1st August 2010
26 July I've received Efika MX.
I've found hdmi-to-dvi cable and euro power cable.
Device was booted successfully into ubuntu 9.04 minimal.
Now I should install kexec-tools and try to kexec'ing some kernels.
P.S. I've written more in russian here: http://www.ul-lug.ru/node/50
|
|
|
|
|
|
|
|
|
|
|
|
Benchmark comparison between the Efika MX and an Asus Eee 701
Project: Port of the Enlightenment Foundation Libraries plus the E17 desktop (Applications & Software)
posted by Luis Felipe Strano Moraes on 29th July 2010
The EFL have a benchmark program called expedite that runs a series of tests to measure rendering speed of the device. Here is a chart comparing the output of running it on the Efika and on an Asus Eee 701 (Celeron 900mhz CPU):
Expedite Comparison Chart
Notice that the EFL that is running on the Efika is not completely optimized yet, and the main difference in this case being that the Eee has SSE2 and other SIMD optimizations enabled, and the Efika version was compiled without NEON support. I'm moving to a new apartment this weekend and am already without internet access at home, but I'll soon try to get the benchmark with NEON enabled to get a better comparison going.
|
|
|
|
|
|
