Hi Matt, This is great, thanks! The Grouper code is actually not used in the CLA. It is an older machine learning algorithm we developed for doing temporal clustering of patterns.
The NuPIC codebase includes a bunch of code from our previous generation of algorithms which we don't use anymore. We had them in our repository mostly for historical reasons. It might be worthwhile doing a serious pass at removing that cruft. I don't know what people feel about removing some of that code - any thoughts? --Subutai On Fri, Aug 16, 2013 at 3:33 PM, Matt Keith <[email protected]> wrote: > Hello all, > > I have been trying to get a build to work on a Raspberry Pi. I have made > some good progress and I thought that I would share my findings and see if > anyone out there has any additional suggestions. The Raspbian (Debian) > distribution seems to give the best performance, but I was able to make > some headway on the Pidora (Fedora) build as well. Here is my current > cookbook for building the code. You can get my source changes from > https://github.com/keithcom/nupic . > > Start with the Raspbian Wheezy image > - sudo apt-get update > - sudo apt-get install automake libtool python-dev swig libapr1-dev > libaprutil1-dev > - install setuptools - https://pypi.python.org/pypi/setuptools/0.9.8 > - install pip - https://pypi.python.org/pypi/pip > - install libyaml - http://pyyaml.org/wiki/LibYAML > - git clone https://github.com/keithcom/nupic.git > - edit .bashrc as per the README.md > - create $NUPIC/external/linux32/lib dir and ln -s > /usr/local/lib/libyaml.a there > - $NUPIC/build.sh and wait > > Let me know if you would like my cookbook for the Pidora image. > > Note that I have added a new section to the > external/common/include/cycle_counter.hpp file to try and use the system > counter since there is no real time clock on the raspberry pi. I still > need to find a define that is specific to the Pi to test for in the file > and I haven't been able to test my code changes yet, but it should be a > good start. > > Right now the code compiles through to the libalgorithms.a and dies on the > Grouper.o with the following error: > {standard input}: Assembler messages: > {standard input}:19864: Error: co-processor offset out of range > make[2]: *** [libalgorithms_a-Grouper.o] Error 1 > > I welcome any and all feedback to help get this working. > > Thanks, > > Matt > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >
_______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
