Running Caffe AlexNet/GoogleNet On Some CPUs Compared To NVIDIA CUDA
Written by Michael Larabel in NVIDIA on 29 August 2016 at 06:25 PM EDT. 6 Comments
NVIDIA --
With working on some Broadwell-EP Linux comparison benchmarks this weekend, as part of that onslaught of benchmarks I decided to run the CPU-only Caffe build on a few different Intel CPUs. For fun, afterwards I checked to see how the performance compares to Caffe with CUDA+cuDNN on a few Maxwell/Pascal GPUs.

For those wondering why GPUs are so heavily favored when it comes to deep learning, the results are obvious.

These tests for fun compared a Xeon E5-2609 v4 Broadwell-EP, Xeon E5-2687W v3 Haswell, and Core i7 5960X Haswell-E to a low-end GeForce GTX 950, GTX TITAN X, and GTX 1080. All tests were done using the same Caffe code-base as packaged via the Phoronix Test Suite and OpenBenchmarking.org. Caffe's GoogleNet and AlexNet tests were run for 200 iterations on each platform.

Just in case you haven't looked at any other CPU vs. GPU comparisons for AlexNet... CUDA 8.0 was in use with the NVIDIA hardware. Even the low-end GeForce GTX 950 is many times faster than the different CPUs tried.

Just some fun benchmarks to pass along. You can easily run your own comparison tests via the Phoronix Test Suite open-source, cross-platform benchmarking software.

About The Author
Author picture

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 10,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter or contacted via MichaelLarabel.com.

Related NVIDIA News
Popular News This Week