MaTEx Performance

Charles Siegel edited this page Mar 28, 2017 · 9 revisions

MaTEx Deep Learning Performance

Experimental Testbed and Software Description:

  1. Eight nodes, each of which has an Intel 20-core Haswell, InfiniBand FDR with OpenMPI 1.8.3, NVIDIA K40m GPUs using CUDA 7.5 and cuDNN 4.

  2. Twenty nodes, each of which has an Intel 20-core IvyBridge, InfiniBand FDR with OpenMPI 1.8.4

Neural Networks

We test AlexNet, GoogLeNet, InceptionV3 and ResNet50 networks.

Results

Scaling of ImageNet Models relative to themselves on a single node, for testbeds 1 and 2 consisting, respectively, of GPUs and CPUs.

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.