Optimal number of threads determination system

A computing device comprising:

• a processor;
• and
• a computer-readable medium operably coupled to the processor, the computer-readable medium having computer-readable instructions stored thereon that, when executed by the processor, cause the computing device to
• control execution of a computing task in a computing environment based on each task configuration included in a plurality of task configurations to determine an execution runtime value for each task configuration,
• wherein each task configuration of the plurality of task configurations includes a task parameter value, a dataset indicator, a hardware indicator, and a number of threads value,
• wherein the task parameter value is an input to the computing task,
• wherein the dataset indicator indicates a dataset to use when executing the computing task,
• wherein the hardware indicator indicates an executing computing device in the computing environment used to execute the computing task,
• wherein the number of threads value indicates a number of threads to use to execute the computing task using the executing computing device;
• determine an optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset indicator, and the hardware indicator, wherein the optimal number of threads value is determined as an extremum value of an execution parameter value as a function of the number of threads value;
• determine a dataset parameter value for the dataset;
• determine a hardware parameter value as a characteristic of each distinct executing computing device in the computing environment indicated using the hardware indicator;
• store, in a performance dataset, the determined optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset parameter value, and the hardware parameter value in association with the common values for the task parameter value, the dataset parameter value, and the hardware parameter value;
• train a thread determination model to determine a number of threads using the stored performance dataset, wherein the thread determination model is a type of machine learning model;
• and
• output the trained thread determination model.

A non-transitory computer-readable medium having stored thereon computer-readable instructions that when executed by a computing device cause the computing device to:

• control execution of a computing task in a computing environment based on each task configuration included in a plurality of task configurations to determine an execution runtime value for each task configuration,
• wherein each task configuration of the plurality of task configurations includes a task parameter value, a dataset indicator, a hardware indicator, and a number of threads value,
• wherein the task parameter value is an input to the computing task,
• wherein the dataset indicator indicates a dataset to use when executing the computing task,
• wherein the hardware indicator indicates an executing computing device in the computing environment used to execute the computing task,
• wherein the number of threads value indicates a number of threads to use to execute the computing task using the executing computing device;
• determine an optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset indicator, and the hardware indicator, wherein the optimal number of threads value is determined as an extremum value of an execution parameter value as a function of the number of threads value;
• determine a dataset parameter value for the dataset;
• determine a hardware parameter value as a characteristic of each distinct executing computing device in the computing environment indicated using the hardware indicator;
• store, in a performance dataset, the determined optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset parameter value, and the hardware parameter value in association with the common values for the task parameter value, the dataset parameter value, and the hardware parameter value;
• train a thread determination model to determine a number of threads using the stored performance dataset, wherein the thread determination model is a type of machine learning model;
• and
• output the trained thread determination model.

A computing device comprising:

• a processor;
• and
• a computer-readable medium operably coupled to the processor, the computer-readable medium having computer-readable instructions stored thereon that, when executed by the processor, cause the computing device to
• control execution of a computing task in a computing environment based on each task configuration included in a plurality of task configurations to determine an execution runtime value for each task configuration,
• wherein the computing task is training a machine learning model,
• wherein each task configuration of the plurality of task configurations includes a task parameter value, a dataset indicator, a hardware indicator, and a number of threads value,
• wherein the task parameter value is an input to the computing task,
• wherein the dataset indicator indicates a dataset to use when executing the computing task,
• wherein the hardware indicator indicates an executing computing device in the computing environment used to execute the computing task,
• wherein the number of threads value indicates a number of threads to use to execute the computing task using the executing computing device;
• determine an optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset indicator, and the hardware indicator, wherein the optimal number of threads value is determined as an extremum value of an execution parameter value as a function of the number of threads value;
• determine a dataset parameter value for the dataset;
• determine a hardware parameter value as a characteristic of each distinct executing computing device in the computing environment indicated using the hardware indicator;
• and
• store, in a performance dataset, the determined optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset parameter value, and the hardware parameter value in association with the common values for the task parameter value, the dataset parameter value, and the hardware parameter value.

A method of determining an optimal number of threads for a computer task, the method comprising:

• controlling, by a computing device, execution of a computing task in a computing environment based on each task configuration included in a plurality of task configurations to determine an execution runtime value for each task configuration,
• wherein the computing task is training a machine learning model,
• wherein each task configuration of the plurality of task configurations includes a task parameter value, a dataset indicator, a hardware indicator, and a number of threads value,
• wherein the task parameter value is an input to the computing task,
• wherein the dataset indicator indicates a dataset to use when executing the computing task,
• wherein the hardware indicator indicates an executing computing device in the computing environment used to execute the computing task,
• wherein the number of threads value indicates a number of threads to use to execute the computing task using the executing computing device;
• determining, by the computing device, an optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset indicator, and the hardware indicator, wherein the optimal number of threads value is determined as an extremum value of an execution parameter value as a function of the number of threads value;
• determining, by the computing device, a dataset parameter value for the dataset;
• determining, by the computing device, a hardware parameter value as a characteristic of each distinct executing computing device in the computing environment indicated using the hardware indicator;
• and
• storing, by the computing device, in a performance dataset, the determined optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset parameter value, and the hardware parameter value in association with the common values for the task parameter value, the dataset parameter value, and the hardware parameter value.

A method of determining an optimal number of threads for a computer task, the method comprising:

• controlling, by a computing device, execution of a computing task in a computing environment based on each task configuration included in a plurality of task configurations to determine an execution runtime value for each task configuration,
• wherein each task configuration of the plurality of task configurations includes a task parameter value, a dataset indicator, a hardware indicator, and a number of threads value,
• wherein the task parameter value is an input to the computing task,
• wherein the dataset indicator indicates a dataset to use when executing the computing task,
• wherein the hardware indicator indicates an executing computing device in the computing environment used to execute the computing task,
• wherein the number of threads value indicates a number of threads to use to execute the computing task using the executing computing device;
• determining, by the computing device, an optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset indicator, and the hardware indicator, wherein the optimal number of threads value is determined as an extremum value of an execution parameter value as a function of the number of threads value;
• determining, by the computing device, a dataset parameter value for the dataset;
• determining, by the computing device, a hardware parameter value as a characteristic of each distinct executing computing device in the computing environment indicated using the hardware indicator;
• storing, by the computing device, in a performance dataset, the determined optimal number of threads value for each set of task configurations in the plurality of task configurations that have common values for the task parameter value, the dataset parameter value, and the hardware parameter value in association with the common values for the task parameter value, the dataset parameter value, and the hardware parameter value;
• training, by the computing device, a thread determination model to determine a number of threads using the stored performance dataset, wherein the thread determination model is a type of machine learning model;
• and
• outputting, by the computing device, the trained thread determination model.