Subject: | |
From: | |
Reply To: | |
Date: | Fri, 22 Apr 2016 10:49:47 +0100 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Hi,
in case anyone ever wondered, I can confirm that the Spikey software runs just fine on a BeagleBone. The BeagleBone is a credit-card sized computer with an ARM processor that can run a full linux system (architecture armv7l), not unlike the Raspberry Pi.
We’re using the BeagleBone because our goal is to come up with a portable, small-footprint neuromorphic system. It could also be interesting for the next-generation dissemination systems.
The software compiled quite fine, with three minor hiccups:
- on ARM ubuntu, the boost libraries are installed in a location that the waf-script doesn’t know about: /usr/lib/arm-linux-gnueabihf
- it ran out of memory when compiling the logger. The beaglebone only comes with 512 MB. Using a temporary swap file solved this (mkswap/swapon).
- I had to use the CPATH environment variable to manually point it to numpy’s arrayobject.h, which was located in an anaconda env (probably rather a problem with anaconda than with the beaglebone per se).
I should say though that compilation took a very long time (might have been an hour or so). This is probably a consequence of a not-so-fast processor, little and slow RAM and slow disk i/o. So some cross-compilation setup may be useful if you’re in a hurry.
Cheers,
Michael
--
Michael Schmuker
University of Sussex
School of Engineering and Informatics
Falmer, Brighton BN1 9QJ
Tel: +44 (0) 1273 876565
http://biomachinelearning.net
########################################################################
To unsubscribe from the KIP-SPIKEY-USERS list, click the following link:
https://listserv.uni-heidelberg.de/cgi-bin/wa?SUBED1=KIP-SPIKEY-USERS
|
|
|