Skip to main content

API Reference

Burla is a python package with only one function (really!). Here is the API-Documentation for this function:

burla.remote_parallel_map​

burla.remoteparallel_map(function, inputs, parallelism=-1, func_cpu=1, func_ram=1, func_gpu=0, verbose=True, image=None)

Runs provided function_ on each item in inputs at the same time, in the cloud, each on a separate CPU, up to 4000 CPUs. If more than 4000 inputs are provided, inputs are queued and processed progressively. When run, expect this function to be in the preparing state for at least 3 minutes.

Parameters:​

  • function_ : Callable

    • Python function. Must have single input argument, eg: function_(inputs[0]) does not raise an exception.
    • Input & return value data types cannot be any of: frame (not DataFrame those are ok!), generator, traceback.

  • inputs : List

    • List containing elements passable to function_.
    • Element data types can be anything except: frame (not DataFrame those are ok!), generator, traceback.

  • parallelism : Int, default: -1

    • Number of calls to function_ running in parallel.
    • Set to -1 for maximum (4000 or len(inputs), whichever is greater).

  • func_cpu : Int, default: 1

    • CPU allocated for every individual call to function_, max 96.

  • func_ram : Int, default: 1

    • RAM allocated for every individual call to function_, max 624.

  • func_gpu : Int, default: 0

    • GPUs (Nvidia Tesla-T4s) allocated for every individual call to function_, max 4.

  • verbose : bool, default: True

    • Optional, verbosity, if False status indicator is not displayed.

  • image : Optional[str], default: None

    • Optional, URI of a publicly accessible docker image. If None, Burla will attempt to copy and install the local environment on all remote machines.

  • packages : Optional[List[str]], default: None

    • Optional, list of pip packages to install on machines where function_ will run, If this is used packages will not be automatically detected and installed.

  • api_key : Optional[str], default: None

    • Optional, api_key, for use in situations where burla is deployed inside backend services, please email jake@burla.dev to have an api-key issued.

Returns:​

  • List :
    • Outputs returned by function_ for every input in inputs, ordering of this list will not match the order of inputs.

We know you may have additional questions about how remote_parallel_map works and we're working hard to add content that may answer these questions.
For now the best we can do is recommend you schedule a call with us, or shoot us an email! We're always happy to talk.