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.
|
Classes | |
class | UniqueMessageFilter |
Prevents the logger from filtering duplicate messages. More... | |
class | DiGraphWrapper |
Wraps the DiGraph class with type annotations for the nodes. More... | |
Functions | |
int | hash_sha512 (Any data) |
Hashes the input data using SHA-512. More... | |
Any | pickle_deepcopy (Any to_copy) |
def | pickle_save (str to_save, str path) |
def | pickle_load (str path) |
dict[str, Any]|list[dict[str, Any]] | open_yaml (str path) |
Any | json_repr_handler (Any obj, bool simple=False) |
Recursively converts objects into a json representation. More... | |
Variables | |
T = TypeVar("T") | |
int zigzag.utils.hash_sha512 | ( | Any | data | ) |
Hashes the input data using SHA-512.
Any zigzag.utils.json_repr_handler | ( | Any | obj, |
bool | simple = False |
||
) |
Recursively converts objects into a json representation.
dict[str, Any] | list[dict[str, Any]] zigzag.utils.open_yaml | ( | str | path | ) |
Any zigzag.utils.pickle_deepcopy | ( | Any | to_copy | ) |
def zigzag.utils.pickle_load | ( | str | path | ) |
def zigzag.utils.pickle_save | ( | str | to_save, |
str | path | ||
) |
T = TypeVar("T") |