SMAUG
Simulating Machine Learning Applications on gem5-Aladdin
|
3 #include "smaug/core/network.h"
4 #include "smaug/core/workspace.h"
5 #include "smaug/core/operator.h"
15 : network(_network), workspace(_workspace) {}
Tensor represents a single multi-dimensional array of data.
Network encapsulates all of the information SMAUG will use during execution: the overall computation ...
void maybeRunOperator(Operator *op)
If none of the inputs to the current Operator are dead, then this will run the Operator; otherwise,...
void updateChildren(Operator *op)
After an Operator is run, this updates the number of pending inputs on all its children.
Workspace is the container and owner of all Tensors and Operators in the Network.
Tensor * runNetwork()
Runs the Network to completion.
Tensor * scheduleReady()
Runs the operators in the ready queue.
Operator is the base class for all graph operators supported by SMAUG.
The smaug namespace is the parent namespace of all C++ code in SMAUG.
Scheduler is responsible for running the Network.
std::list< Operator * > readyQueue
The queue of all Operators ready to be executed.