Skip to content

Implementation of 2D Convolution operation for Neural Networks using Intel x86(i368)/x86-6(amd64) AVX-256 instructions. All data flow methods, i.e input stationary, weight stationary and output stationary are implemented. The forward pass of Alexnet architecture is constructed using it.

Sooryakiran/Convolution-with-AVX

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Convolutional Neural Networks implimentation using intel AVX instructions

CS 6886: Systems Engineering for Deep Learning

Assignment 2 Submitted by: Sooryakiran P ME17B174 sooryakiran@smail.iitm.ac.in


To compile everything, cd src make all

To run dummy inference on alexnet once, cd src && ./alexnet

To run dummy inference on alexnet 100 times, cs src && ./alexnet_100


Directory Structure

.
├── Figures		     // Directory containing all figures
│   └── ...	
│	
├── ME17B174_A2.pdf	     // Copy of report
├── Question.pdf	     // Assignment Questions
│
├── README.md		     // Goto line 1
├── data		     // Directory containing data used for
│			        plotting
├── report.odt		     // Editable report
└── src
    ├── Makefile	     // Makefile for compilation
    ├── TODO		     // TODO
    ├── alexnet.cpp	     // Alexnet source code
    ├── alexnet_100.cpp	     // Alexnet inf. 100 times source code
    ├── test.cpp	     // Source code for testind individual
    │				layers
    ├── util.cpp	     // Main implimentation of layers
    └── util.h		     // Header file for layer definitions

=================================================================

About

Implementation of 2D Convolution operation for Neural Networks using Intel x86(i368)/x86-6(amd64) AVX-256 instructions. All data flow methods, i.e input stationary, weight stationary and output stationary are implemented. The forward pass of Alexnet architecture is constructed using it.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published