qat.qlmaas.plugins.QLMaaSPlugin

Remote Plugins are managed by QLMaaSPlugin objects. These objects can be built using:

• qlmaas.plugins module if a configuration exists on your computer

• a QLMaaSConnection object (when the connection is defined manually). Please note that, even if there is no configuration file on your computer, instantiating a QLMaaSConnection object enables the qlmaas.plugins module

class qat.qlmaas.plugins.QLMaaSPlugin(connection, name: str, constructor: SerializedArguments, description: str)

Class describing a Plugin defined on a remote server.

If a configuration file is present on your computer, you can use the module qlmaas.plugins to instantiate a remote Plugin:

from qlmaas.plugins import Nnizer

# Define a plugin
plugin = Nnizer()

Otherwise, if no configuration file is set on your machine, you can use a QLMaaSConnection object:

from qat.qlmaas import QLMaaSConnection

# Connect to the server
conn = QLMaaSConnection()

# Get remote Plugin
Nnizer = conn.get_qpu("qat.plugins:Nnizer")
plugin = Nnizer()

compile(batch, specs, meta_data: Optional[dict] = None, mem_usage: Optional[int] = None, core_usage: Optional[int] = None, nb_nodes: Optional[int] = None)

Compile a batch

Parameters

• batch (Batch) – batch to compile

• specs (HardwareSpecs) – specs

• meta_data (dict, optional) – override the meta_data of the batch

• mem_usage (int, optional) – override default memory usage (in MB) calculation

• core_usage (int, optional) – override default core usage calculation

Returns

asynchronous result

Return type

AsyncResult

compile_job(*args, **kwargs)

Compiles a single job

Parameters

• job (Job) – job to compile

• specs (HardwareSpecs) – specs

• meta_data (dict, optional) – override the meta_data of the batch

• mem_usage (int, optional) – memory usage (in MB)

• core_usage (int, optional) – core usage

Returns

asynchronous result

Return type

AsyncResult