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 | Namespaces | Variables
data_movement.py File Reference

Classes

class  FourWayDataMoving
 Represents a standard four-way data moving attribute of a memory interface. More...
 
class  MemoryAccesses
 Represents the number of memory accesses in four directions. More...
 
class  AccessEnergy
 Represents the memory access energy in four directions. More...
 
class  DataMoveAttr
 
class  DataMovePattern
 Collects the memory access pattern for each unit memory (memory holding one operand at one level). More...
 

Namespaces

 zigzag.mapping.data_movement
 

Variables

 T = TypeVar("T", bound=int | float)