AI Architecture & Systems
1
Microarchitecture
2
Kernel
3
Compiler
4
Architecture
5
System
6
Algorithms
Compiler
/
§3.5
Graph-Level Optimization
3.5.1
Constant Folding, CSE, and Dead Code Elimination
#
3.5.2
Operator Fusion, Operator Decomposition, and Pattern Rewriting
#
3.5.3
Algebraic Simplification and Operator Reordering
#
3.5.4
Attention, MLP, and Transformer-Specific Fusion
#
3.5.5
Numerical Equivalence and Optimization Legality
#
← Previous
3.4 Automatic Differentiation and Training-Graph Compilation
→ Next
3.6 Loop Transformation and Tensorization