turi::neural_net::TaskQueue Class Reference

Abstract task queue interface modeled after Grand Central Dispatch. More...

#include <ml/neural_net/TaskQueue.hpp>

Public Member Functions
virtual void	DispatchAsync (std::function< void()> task)=0
virtual void	DispatchSync (std::function< void()> task)=0
virtual void	DispatchApply (size_t n, std::function< void(size_t i)> task)=0

Static Public Member Functions
static std::shared_ptr< TaskQueue >	GetGlobalConcurrentQueue ()
static std::unique_ptr< TaskQueue >	CreateSerialQueue (const char *label)

Detailed Description

Abstract task queue interface modeled after Grand Central Dispatch.

Definition at line 17 of file TaskQueue.hpp.

Member Function Documentation

CreateSerialQueue()

static std::unique_ptr<TaskQueue> turi::neural_net::TaskQueue::CreateSerialQueue ( const char *  label )

static

Returns a task queue that guarantees that if task A is submitted before task B, then task A will finish before task B begins. Accepts a label that may be used by the system to identify work done by this queue.

DispatchApply()

virtual void turi::neural_net::TaskQueue::DispatchApply ( size_t  n, std::function< void(size_t i)>  task  )

pure virtual

Submits a function to this task queue n times, with arguments ranging from 0 to n - 1. When dispatched to a concurrent queue, the function must be reentrant. Rethrows the first exception thrown by any task invocation.

Implemented in turi::neural_net::GlobalPosixTaskQueue, turi::neural_net::SerialPosixTaskQueue, and turi::neural_net::GrandCentralDispatchQueue.

DispatchAsync()

virtual void turi::neural_net::TaskQueue::DispatchAsync ( std::function< void()>  task )

pure virtual

Submits a function to this task queue without waiting for the function to finish. The task must not throw an exception.

Implemented in turi::neural_net::GrandCentralDispatchQueue, and turi::neural_net::PosixTaskQueue.

DispatchSync()

virtual void turi::neural_net::TaskQueue::DispatchSync ( std::function< void()>  task )

pure virtual

Submits a function to this task queue and waits for the function to execute. The task must not throw an exception.

Implemented in turi::neural_net::GrandCentralDispatchQueue, and turi::neural_net::PosixTaskQueue.

GetGlobalConcurrentQueue()

static std::shared_ptr<TaskQueue> turi::neural_net::TaskQueue::GetGlobalConcurrentQueue ( )

static

Returns a task queue that does not enforce any ordering among its tasks and that shares system resources with other task queues created by this function.

---

The documentation for this class was generated from the following file:

• ml/neural_net/TaskQueue.hpp