Amazon Web Services (AWS) and
Intel share a passion for delivering constant innovation on
the cloud. Together, they have developed a avriety of
resources and technologies for High Performance Computing,
Big Data, Artificial Intelligence / Machine Learning and
the Internet of Things.
Unlock the Potential of Your AWS
Instance with Intel Parallel Studio XE
Optimized Deep Learning Frameworks on EC2 CPU Instances
AWS DL AMI v6.0 includes
Intel® Optimizations for TensorFlow* - pre-built and
ready to use with Intel® Math Kernel Library and Math
Kernel Library for Deep Neural Networks (Intel® MKL and
Intel® MKL-DNN) primitives.
For the data scientist,
consequences of using Intel Optimization for TensorFlow
manifest as significant speedups across the deep learning
design space. Figure 1 xompares default TensorFlow
(previously available in AWS' DL AMI V5.0) vs. Intel
Optimization for TensorFlow (currently available in AWS' DL
AMI V6.0) training throughput on four benchmark topologies:
InceptionV3, ResNet50, ResNet152 and VGG16.