Azure AutoML uses only 1 core RRS feed

  • Question

  • Hi,

    I'm testing Azure AutoML in local mode, and only 1 core is used on my workstation (same on my server).

    no matter the max concurrent settings I put in place.

    so there is no parallel operation executed, no multi-threading.

    at least I'm faster than Azure nodes at fraction of the cost, but still surprised how badly optimized this API is compared to some R tools.(So I presume that an Azure VM is also under used... is it setup to cost us a lot by paying for big under used VM?)

    is there any special stuff to know I should setup on my PC? 

    Tuesday, November 5, 2019 1:42 PM

All replies

  • Hello Willgart1,

    You can use the setting max_cores_per_iteration to control the number of cores that would be available for AutoML to use. You can set it to -1 to use maximum available. You can find the setting in documentation here.

    Also, training multiple iterations or models in parallel is supported only in remote runs (I.e using AML compute etc)


    Thursday, November 7, 2019 10:37 AM
  • Hi,

    Sorry, but it doesnt works.

    the max cores option did not change anything.

    maybe there is a missing library?

    I'll conduct additional tests today.

    Thursday, November 7, 2019 12:07 PM