The Linley Processor Conference that wrapped up last week has historically has been around processor development, and this year's conference was the same but with a twist. Many of the processor developments were framed around, or had something to do with Machine Learning. IP vendor's presentations focused on machine learning. Processors had some message about machine learning or acceleration of workloads.
Education on Alexnet, VGG16, Resnet and Yolo were interesting to learn and tied very well into the Micron demonstration on Wednesday evening. The Yolo image above is a recognition application running on an Nvidia GEForce 1080Ti graphics card which leverages the Micron GDDR5X devices. This was 11GBytes of capacity running 10-20 frames per second using the "YOLOv2 608x608" model, which created a great discussion point with many attendees of the conference. More importantly, it supported the message of proven technology leading continuing the high-performance memory roadmap, leading to our GDDR6N announcement.
If you are interested in seeing more about what Micron is doing with Machine Learning, come see us at SC17 November 13-16 in Denver, where we will provide more demonstrations on active systems and memory requirements for machine learning applications.