Cyclops Tensor Framework
parallel arithmetic on multidimensional arrays
|
#include <schedule.h>
Public Member Functions | |
Schedule (World *world=NULL) | |
Constructor, optionally specifying a world to restrict processor allocations to. More... | |
void | record () |
Starts recording all tensor operations to this schedule (instead of executing them immediately) More... | |
ScheduleTimer | execute () |
Executes the schedule and implicitly terminates recording. More... | |
ScheduleTimer | partition_and_execute () |
Executes a slide of the ready_queue, partitioning it among the processors in the grid. More... | |
void | schedule_op_successors (TensorOperation *op) |
Call when a tensor op finishes, this adds newly enabled ops to the ready queue. More... | |
void | add_operation_typed (TensorOperation *op) |
Adds a tensor operation to this schedule. THIS IS CALL ORDER DEPENDENT - operations will appear to execute sequentially in the order they were added. More... | |
void | add_operation (TensorOperationBase *op) |
void | set_max_partitions (int in_partitions) |
Protected Attributes | |
World * | world |
std::deque< TensorOperation * > | root_tasks |
std::deque< TensorOperation * > | steps_original |
std::map< CTF_int::tensor *, TensorOperation * > | latest_write |
std::deque< TensorOperation * > | ready_tasks |
int | partitions |
Definition at line 138 of file schedule.h.