accelerator_view Class
Represents a virtual device abstraction on a C++ AMP data-parallel accelerator.
class accelerator_view;
Members
Public Constructors
Name |
Description |
---|---|
Initializes a new instance of the accelerator_view class. |
|
Destroys the accelerator_view object. |
Public Methods
Name |
Description |
---|---|
Returns a future to track the completion of all commands submitted so far to this accelerator_view object. |
|
Submits all pending commands queued to the accelerator_view object to the accelerator for execution. |
|
Returns the accelerator object for the accelerator_view object. |
|
Returns a Boolean value that indicates whether the accelerator_view object has the DEBUG layer enabled for extensive error reporting. |
|
Returns the queuing mode for the accelerator_view object. |
|
Returns the version of the accelerator_view. |
|
Waits for all commands submitted to the accelerator_view object to finish. |
Public Operators
Name |
Description |
---|---|
Compares this accelerator_view object with another and returns false if they are the same; otherwise, returns true. |
|
Copies the contents of the specified accelerator_view object into this one. |
|
Compares this accelerator_view object with another and returns true if they are the same; otherwise, returns false. |
Public Data Members
Name |
Description |
---|---|
Gets the accelerator object for the accelerator_view object. |
|
Gets a Boolean value that indicates whether the accelerator_view object has the DEBUG layer enabled for extensive error reporting. |
|
Gets the queuing mode for the accelerator_view object. |
|
Gets the version of the accelerator. |
Inheritance Hierarchy
accelerator_view
Remarks
An accelerator_view object represents a logical, isolated view of an accelerator. A single physical compute device can have many logical, isolated accelerator_view objects. Each accelerator has a default accelerator_view object. Additional accelerator_view objects can be created.
Physical devices can be shared among many client threads. Client threads can cooperatively use the same accelerator_view object of an accelerator, or each client can communicate with a compute device via an independent accelerator_view object for isolation from other client threads.
An accelerator_view object can have one of two queuing_mode Enumeration states. If the queuing mode is immediate, commands like copy and parallel_for_each are sent to the corresponding accelerator device as soon as they return to the caller. If the queuing mode is deferred, such commands are queued up on a command queue that corresponds to the accelerator_view object. Commands are not actually sent to the device until flush() is called.
Requirements
Header: amprt.h
Namespace: Concurrency