Our DGX-1 Workshop Launches
April 30, 2019 – Organizations are increasingly coming to the realization that they need AI at scale and enterprise IT teams are increasingly taking an interest in innovations that will enable them to fulfill this goal. NVIDIA’s AI supercomputer DGX-1 created quite the buzz when it launched last year and quickly became the AI compute standard for enterprise. Offering the fastest integrated system for deep learning, the DGX-1 lets data scientists uncover patterns in large data sets that lead to new knowledge and insights in hours or minutes instead of weeks and months. It is architected for high throughput and high interconnect bandwidth to maximize neural network training performance.
Given the interest of our customers in learning more about the DGX-1, last week we launched our first DGX-1 workshop to great enthusiasm. Led by our Senior Solutions Architect for Data & AI, Patrick Donnelly, the three-hour session introduced the DGX-1 system and containerized deep learning applications and covered a lot of valuable information including:
- Review of the DGX-1 hardware and software stack
- Commands for configuring NFS, InfiniBand, system profiling and health monitoring
- Docker and NVIDIA container runtimes for Docker
- An overview of the operations behind training a convolutional neural network
- Using the NVIDIA GPU Cloud PyTorch container and profiling the code using nvprof
- Using DeepOps to set up and run jobs on a Kubernetes cluster
- A tour of our POC lab, integration lab and DGX POD
A Three Day Class in a Three Hour Workshop
The workshop was lauded for its interactivity and educational usefulness for the participants, who found the session to be invaluable in their continuing deep learning journey. One of the attendees commented that he had learned more in the 3-hour workshop in helping him to understand the infrastructure and algorithms necessary to run a deep learning job than he had in the last few years of attempting to learn it all on his own. Another participant noted that he too had absorbed a lot of information and praised Patrick for his knowledge and talent in being able to pack the equivalent of a 3-day class into just a few hours.
With the positive feedback and learning opportunities for our customers, we look forward to hosting future sessions. We offer variety of workshops that are designed to enhance the participants’ knowledge of emergent technologies with training that can be customized to fit their needs. For more information visit our website page: https://www.groupwaretech.com/workshops/Back To All