ARM based installation
bwees opened this issue · 5 comments
I am setting up a raspberry pi to run my algorithms on. I am creating my own Docker Image for the raspberry pi that is based on the provided dockerfiles. I am having issues with the pip install with the error:
numpy.distutils.system_info.NotFoundError: No lapack/blas resources found.
If anyone has successfully made a docker image for ARM, please let me know
Thnaks
sounds cool, do update us if you find a solution
I have successfully created and ARM-based docker image of pylivetrader. If you would like to use, run the following:
docker pull bwees/pylivetrader:latest
Caveats include:
- Python 3.7 vs 3.6
- Pipeline-Live is not installed due to the need for compiling zipline
- Pandas 0.23 is used because 0.22 is not compiled in 3.7 and the only need for 0.22 is because of zipline
I will keep this issue open if it interests other users.
out of curiosity - what are some of the advantages you could point out, running pylivetrader on a raspberry pi?
The Raspberry Pi is super low powered so you can run small to medium sized algorithms without having to use a larger desktop or server. The performance is nowhere near a modern day x86 computer but, for most server and number crunching tasks, it performs adequately.
I personally have 2 Raspberry Pi 3B+ (Originally 4 running BOINC COVID research) in a Docker Swarm configuration so it has both failover and distributed workload capabilities. Pylivetrader runs in tandem with some other services on the cluster such as network-wide ad-blocking and a VPN server. I have tested my algorithm which uses about 35 Tech Indicators for decisions and it can process decisions for 25-30 stocks every minute. The main reason I use it is the fact that its not a desktop sucking a large amount of power from the wall and being noisy. The Raspberry Pi is only $35 so it makes it perfect for server applications.