ZigZag - Deep Learning Hardware Design Space Exploration
This repository presents the novel version of our tried-and-tested hardware Architecture-Mapping Design Space Exploration (DSE) Framework for Deep Learning (DL) accelerators. ZigZag bridges the gap between algorithmic DL decisions and their acceleration cost on specialized accelerators through a fast and accurate hardware cost estimation.
|
Class that dumps all received CMEs into a list and saves that list to a pickle file. More...
Public Member Functions | |
def | __init__ (self, list[StageCallable] list_of_callables, *str pickle_filename, **Any kwargs) |
def | run (self) |
Run the simple save stage by running the substage and saving the CostModelEvaluation simple json representation. More... | |
![]() | |
def | __init__ (self, list["StageCallable"] list_of_callables, **Any kwargs) |
def | __iter__ (self) |
bool | is_leaf (self) |
Public Attributes | |
pickle_filename | |
![]() | |
kwargs | |
list_of_callables | |
Class that dumps all received CMEs into a list and saves that list to a pickle file.
def __init__ | ( | self, | |
list[StageCallable] | list_of_callables, | ||
*str | pickle_filename, | ||
**Any | kwargs | ||
) |
list_of_callables | see Stage |
pickle_filename | output pickle filename |
kwargs | any kwargs, passed on to substages |
def run | ( | self | ) |
Run the simple save stage by running the substage and saving the CostModelEvaluation simple json representation.
This should be placed above a ReduceStage such as the SumStage, as we assume the list of CMEs is passed as extra_info
Reimplemented from Stage.
pickle_filename |