Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Picture for David Gregg

Low precision logarithmic number systems: Beyond base-2


Feb 12, 2021
Syed Asad Alam, James Garland, David Gregg

* 22 pages, 12 figures, 8 tables, conference extension 

  Access Paper or Ask Questions

HOBFLOPS CNNs: Hardware Optimized Bitsliced Floating-Point Operations Convolutional Neural Networks


Jul 11, 2020
James Garland, David Gregg

* 20 pages, 2 tables, 18 figures 

  Access Paper or Ask Questions

Exploiting Weight Redundancy in CNNs: Beyond Pruning and Quantization


Jun 22, 2020
Yuan Wen, David Gregg


  Access Paper or Ask Questions

TASO: Time and Space Optimization for Memory-Constrained DNN Inference


May 21, 2020
Yuan Wen, Andrew Anderson, Valentin Radu, Michael F. P. O'Boyle, David Gregg


  Access Paper or Ask Questions

Composition of Saliency Metrics for Channel Pruning with a Myopic Oracle


Apr 03, 2020
Kaveena Persand, Andrew Anderson, David Gregg


  Access Paper or Ask Questions

Performance-Oriented Neural Architecture Search


Jan 09, 2020
Andrew Anderson, Jing Su, Rozenn Dahyot, David Gregg

* The 2019 International Conference on High Performance Computing & Simulation 

  Access Paper or Ask Questions

A Taxonomy of Channel Pruning Signals in CNNs


Jun 11, 2019
Kaveena Persand, Andrew Anderson, David Gregg


  Access Paper or Ask Questions

Winograd Convolution for DNNs: Beyond linear polinomials


May 13, 2019
Barbara Barabasz, David Gregg


  Access Paper or Ask Questions

Optimal DNN Primitive Selection with Partitioned Boolean Quadratic Programming


Nov 02, 2018
Andrew Anderson, David Gregg


  Access Paper or Ask Questions

Scalar Arithmetic Multiple Data: Customizable Precision for Deep Neural Networks


Sep 27, 2018
Andrew Anderson, David Gregg


  Access Paper or Ask Questions

Error Analysis and Improving the Accuracy of Winograd Convolution for Deep Neural Networks


Sep 22, 2018
Barbara Barabasz, Andrew Anderson, David Gregg


  Access Paper or Ask Questions

Low-memory GEMM-based convolution algorithms for deep neural networks


Sep 08, 2017
Andrew Anderson, Aravind Vasudevan, Cormac Keane, David Gregg

* 13 pages, 16 figures and 3 tables. arXiv admin note: text overlap with arXiv:1704.04428 

  Access Paper or Ask Questions

Parallel Multi Channel Convolution using General Matrix Multiplication


Jul 03, 2017
Aravind Vasudevan, Andrew Anderson, David Gregg

* Camera ready version to be published at ASAP 2017 - The 28th Annual IEEE International Conference on Application-specific Systems, Architectures and Processors. 6 pages 

  Access Paper or Ask Questions

Low Complexity Multiply Accumulate Unit for Weight-Sharing Convolutional Neural Networks


Jan 19, 2017
James Garland, David Gregg

* 4 pages 

  Access Paper or Ask Questions

Spectral Convolution Networks


Nov 16, 2016
Maria Francesca, Arthur Hughes, David Gregg


  Access Paper or Ask Questions