PPoPP 2016
Sat 12 - Wed 16 March 2016 Barcelona, Spain
Mon 14 Mar 2016 10:00 - 10:25 at Mallorca+Menorca - Applications Chair(s): Albert Cohen

Deep neural networks (DNN) have recently achieved extraordinary results in domains like computer vision and speech recognition. An essential element for this success has been the introduction of high performance computing (HPC) techniques in the critical step of training the neural network. This paper describes the implementation and analysis of a network-agnostic and convergence-invariant coarse-grain parallelization of the DNN training algorithm. The coarse-grain parallelization is achieved through the exploitation of the batch-level parallelism. This strategy is independent from the support of specialized and optimized libraries. Therefore, the optimization is immediately available for accelerating the DNN training. The proposal is compatible with multi-GPU execution without altering the algorithm convergence rate. The parallelization has been implemented in Caffe, a state-of-the-art DNN framework. The paper describes the code transformations for the parallelization and we also identify the limiting performance factors of the approach. We show competitive performance results for two state-of-the-art computer vision datasets, MNIST and CIFAR-10. In particular, on a 16-core Xeon E5-2667v2 at 3.30GHz we observe speedups of 8× over the sequential execution, at similar performance levels of those obtained by the GPU optimized Caffe version in a NVIDIA K40 GPU.

Mon 14 Mar

Displayed time zone: Belfast change

10:00 - 11:15
ApplicationsMain conference at Mallorca+Menorca
Chair(s): Albert Cohen INRIA
10:00
25m
Talk
Coarse Grain Parallelization of Deep Neural Networks
Main conference
Link to publication DOI
10:25
25m
Talk
High Performance Model Based Image Reconstruction
Main conference
Xiao Wang Purdue University, USA, Amit Sabne School of Electrical and Computer Engineering, Purdue University, Sherman Kisner High Performance Imaging LLC, Anand Raghunathan School of Electrical and Computer Engineering, Purdue University, Charles Bouman School of Electrical and Computer Engineering, Purdue University, Samuel Midkiff School of Electrical and Computer Engineering, Purdue University
Link to publication DOI
10:50
25m
Talk
Exploiting Accelerators for Efficient High Dimensional Similarity Search
Main conference
Link to publication DOI