Abstract: An example computing device comprises a first processing unit having first capabilities, a second processing unit having second capabilities, and a shared memory accessible by the first processing unit and the second processing unit. The shared memory stores data objects in association with type information indicating the data type of the data objects. The example computing device further comprises an instruction set to, when executed by a processing unit of the computing device, select one of the first processing unit and the second processing unit to perform a computation of a particular type, using data of a particular type stored in the shared memory, wherein the selection is performed based on a predefined affinity of the first processing unit for the particular computation type and/or the particular data type and a predefined affinity of the second processing unit for the particular computation type and/or the particular data type.

Abstract: Examples analyze source code of a task prior to compiling the source code to determine a static property of the task. Examples determine a category for the task based at least in part on the static property. Examples compile the source code to generate a binary of the task. Examples determine execution parameters for the task based at least in part on the category. Examples schedule the binary for execution based at least in part on the execution parameters.

Abstract: An example computing device comprises a first processing unit having first capabilities, a second processing unit having second capabilities, and a shared memory accessible by the first processing unit and the second processing unit. The shared memory stores data objects in association with type information indicating the data type of the data objects. The example computing device further comprises an instruction set to, when executed by a processing unit of the computing device, select one of the first processing unit and the second processing unit to perform a computation of a particular type, using data of a particular type stored in the shared memory, wherein the selection is performed based on a predefined affinity of the first processing unit for the particular computation type and/or the particular data type and a predefined affinity of the second processing unit for the particular computation type and/or the particular data type.

Abstract: Examples analyze source code of a task prior to compiling the source code to determine a static property of the task. Examples determine a category for the task based at least in part on the static property. Examples compile the source code to generate a binary of the task. Examples determine execution parameters for the task based at least in part on the category. Examples schedule the binary for execution based at least in part on the execution parameters.