plotter-5-complete.scd analysis very slow on Apple M1

[josedeosorno]

Hello,

I have tested @tedmoore "Coding a 2D Corpus Explorer" tutorials on a MacBook Pro 16" M1 Max, thanks Ted !!
I have compiled both SuperCollider and FluCoMa for M1 architecture. In the exemple plotter-5-complete.scd I have a similar problem as in Arch Linux https://github.com/flucoma/flucoma-sc/issues/135 when evaluating the "analysis" part, the program stuck and run very slow compared to the video tutorial. In the end it works but it's take a long time...
I have also tried to run the SC server on TCP mode like @elgiano said but it's the same, nothing changes. I guess it's a problem with the M1 version of SuperCollider or with my computer configuration because I have the same problem when loading a large number of buffers in parallel. Unfortunately I don't have another Apple M1 to test.

I wonder if anyone has access to an Apple M1 to test if they have the same problem?

[tedmoore]

Hi @josedeosorno, thanks for the report! I think my colleague @jamesb93 has an M1 he can test on. If any epiphanies or changes happen please let us know!

-T

[josedeosorno]

Hi @tedmoore,
I just compiled a new version of SC for M1 and downloaded the nightly version of flucoma (Universal) https://github.com/flucoma/flucoma-sc/releases/tag/nightly
When I do the analysis it works much faster than the previous version I had compiled. It is still much slower than the x64 intel version but it is getting better 👍 I don't have access to another M1 or M2 computer to test yet...

Thanks again for this excellent library!

Best,

José

[tedmoore]

Hi @josedeosorno,

Glad to hear it's working! Sorry to say I still don't have an M1 or 2 for testing. I hope it's working well enough for you to make some sounds. Let me know if you have any questions.

Thanks for taking the time to do the tests and report back. I hope others will find this useful. And if anyone can test on another M1 and/or M2 and report back, that'd be great.

Cheers,

T