13-Way NVIDIA GeForce + CUDA 8.0 + cuDNN Caffe Benchmarks

[phoronix]

Phoronix: 13-Way NVIDIA GeForce + CUDA 8.0 + cuDNN Caffe Benchmarks

Not that you would normally buy a cheap NVIDIA GeForce graphics card for deep learning tasks like the recently launched GTX 1050 series, as part of running some other fresh CUDA+OpenCL benchmarks I realized I hadn't run any Caffe benchmarks in a while so here are some fresh numbers today. With thirteen NVIDIA GeForce graphics cards including all the consumer GeForce GTX 1000 Pascal cards to date, here are some Caffe benchmarks using the latest NVIDIA 375.10 Linux driver on Ubuntu along with CUDA 8.0 and cuDNN.

http://www.phoronix.com/vr.php?view=23722

[Jumbotron]

It really is amazing to see the Pascal series power consumption numbers in all the tests that Michael runs. I remember the days of the "Hoover Vacuum" sound that Geforce cards used to make in comparison to Radeon cards.

On another note.....I don't think you have any apologies to make Michael about running what amounts to deep learning benchmarks on these cards. All data points from your various benchmarks help all of us factor in what is important to us about spending hard earn cash on these GPU's. It's not like most of us are rolling in cash....either American Greenbacks or Euros. Plus.....this test on top of the others seals the deal for me in upgrading my ancient EVGA GTX 285 Fermi to a GTX 1050 Ti.

[coristus]

Thank you for doing this !. I was just about to make a wrong decision. 1070 it is.