SDAA175
November 2025
AM62A3
,
AM62A7
,
AM67A
,
TDA4VM
1
Abstract
Trademarks
1
Introduction
2
C7xMMA Cache Structure
3
Model DDR Read/Write Analysis for a Compiled TIDL Model
4
Model Optimization
4.1
Simple Structure Models
4.2
Complex Structure
4.2.1
Residual Structures
4.2.2
Parallel Branch Merge
5
Summary
6
References
Application Note
Optimizing TI Deep Learning Performance via Memory and DDR Bandwidth Reduction