python.DeviceConfigFlags Namespace Reference

Classes
class	DeviceBackend

Functions
	createDeviceConfigFlags ()

Function Documentation

createDeviceConfigFlags()

createDeviceConfigFlags

(

)

Function to create the flags for using accelerator devices in Athena

Definition at line 27 of file DeviceConfigFlags.py.

def createDeviceConfigFlags():
    '''Function to create the flags for using accelerator devices in Athena
    '''
 
    # Create the flags container.
    result = AthConfigFlags()
 
    # Memory (resource) management flags.
    result.addFlag('Device.Memory.Cache', True,
                   help='Whether or not to cache memory allocations')
    result.addFlag('Device.Memory.Debug', False,
                   help='Whether or not to actively debug memory allocations')
    result.addFlag('Device.Memory.Monitor', False,
                   help='Whether or not to monitor memory allocations')
    result.addFlag('Device.Memory.Shared', False,
                   help='Whether or not to use "shared" memory for device '
                        'allocations. In case it is set to False, a separate '
                        'host and device memory resource is to be used.')
 
    # Memory copy management flag(s).
    result.addFlag('Device.Copy.Async', True,
                   help='Whether or not to use asynchronous memory copies')
 
    # Backend selection flag(s).
    result.addFlag('Device.Backend', DeviceBackend.CUDA, type=DeviceBackend,
                   help='The backend to use for accelerator devices')
 
    # Return the container.
    return result