Class that handles optimization of temporal mapping given a: More...

Public Member Functions
def	__init__ (self, Accelerator accelerator, LayerNode layer, SpatialMappingInternal spatial_mapping, TemporalMappingType mapping_type, *Any kwargs)
	Initialize the engine with the given: More...

def	run (self, Queue cme_queue)
	Call the necessary methods, start the processes and collect the best temporal mapping found during the run. More...

def	run_simulated_annealing_opt (self, cme_queue)
	Run a simulated annealing optimization on the loop ordering using a loma memory allocation strategy. More...

def	get_temporal_loops (self)
	Get all loops that have to be temporally scheduled given layer and spatial mapping. More...

def	get_prime_factors (self)
	Get the prime factors for all temporal loops in the following format: [('C', 2), ('OY', 2), ('OX', 2), ('K', 7), ...]. More...

Public Attributes
	accelerator

	layer

	spatial_mapping

	mapping_type

	iteration_number

	start_temperature

	opt_criterion_name

	lpf_limit

	cme_queue

	temporal_loop_dim_size

	temporal_mapping_lpf

Detailed Description

Class that handles optimization of temporal mapping given a:

layer
spatial mapping
memory hierarchy
number of iterations
start temperature This optimization is carried out through simulated annealing loop order based. Each loop is broken down to the smallest possible part (prime factors), then a runtime estimation is performed to choose the fastest engine to use (LOMA or SALSA).
TODO cleanup

Constructor & Destructor Documentation

◆ init()

def __init__	(		self,
		*Accelerator	accelerator,
		LayerNode	layer,
		SpatialMappingInternal	spatial_mapping,
		TemporalMappingType	mapping_type,
		**Any	kwargs
	)

Initialize the engine with the given:

LayerNode
SpatialMapping
Accelerator
Number of iterations
Start temperature The memory hierarchy from the correct core is extracted from the accelerator.

Member Function Documentation

◆ get_prime_factors()

def get_prime_factors ( self )

Get the prime factors for all temporal loops in the following format: [('C', 2), ('OY', 2), ('OX', 2), ('K', 7), ...].

◆ get_temporal_loops()

def get_temporal_loops ( self )

Get all loops that have to be temporally scheduled given layer and spatial mapping.

◆ run()

def run	(		self,
		Queue	cme_queue
	)

Call the necessary methods, start the processes and collect the best temporal mapping found during the run.

Here is the caller graph for this function:

◆ run_simulated_annealing_opt()

def run_simulated_annealing_opt	(	self,
		cme_queue
	)

Run a simulated annealing optimization on the loop ordering using a loma memory allocation strategy.

Member Data Documentation

◆ accelerator

accelerator

◆ cme_queue

cme_queue

◆ iteration_number

iteration_number

◆ layer

layer

◆ lpf_limit

lpf_limit

◆ mapping_type

mapping_type

◆ opt_criterion_name

opt_criterion_name

◆ spatial_mapping

spatial_mapping

◆ start_temperature

start_temperature

◆ temporal_loop_dim_size

temporal_loop_dim_size

◆ temporal_mapping_lpf

temporal_mapping_lpf

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

/home/runner/work/zigzag/zigzag/zigzag/opt/salsa/engine.py

Public Member Functions

Public Attributes