I have been running into issues when installing songbird through qiime2. On a previous computer, I was successful with all standard installation and usage. My new computer is running an M1 chip (MacBook Air (M1, 2020)) and wants me to use zsh rather than bash and I thought maybe that could be a problem? I have tried removing al qiime2 environments from conda and reinstalling them but still am met with the error:
QIIME is caching your current deployment for improved performance. This may take a few moments and should only happen once per deployment.
Illegal instruction: 4
Thank you so much for your help!
Hi @ngrayson - I'm not a developer on the songbird plugin (but I am one of the core QIIME 2 developers) - it looks like songbird uses a third-party library/tool called tensorflow - the version of tensorflow it uses might not be compatible with the Apple M1 chip - the error message you shared certainly would support that theory - if tensorflow is sending an instruction that the M1 chip doesn't know how to handle, that message would make sense to me. I'm not too sure what you can do to get around this on this particular machine. In general, I haven't seen too many scientific/academic tools get fully ported over to the M1 chip - I'm actually surprised that you were able to install the QIIME 2 core distribution in the first place.
Hi @thermokarst
Thank you so much for your help. Before buying the computer I checked that qiime2 was compatible but did not think to check any plugins. I am pretty sure that you are right about the version of tensorflow not being supported.
Where did you read about QIIME 2 being compatible with the M1 chip? We haven't done any development to support that formally yet - maybe it was from an unofficial source?
