At SC17 this year we had the opportunity to talk HPC with Dell EMC. Last year most of our discussion with the Dell team centered around traditional HPC applications such as Ansys simulations. This year, there was a noted change: deep learning and AI were a central focus.
What to Expect from Dell EMC Deep Learning in 2018
One item that was previewed is that Dell EMC is working on helping companies in their deep learning / AI journey. While many server makers are happy to offer hardware, there is a dearth of data scientists out there which means taking time to setup and tune systems is not the most prudent use of time for companies. This is the same driver that pushes NVIDIA’s GPU Cloud work and work with its DGX-1. Making the setup steps easier.
We pressed the Dell EMC team a bit, and the service offering will include more than simply setting up Caffe2, mxnet, or Tensorflow on a machine. Instead, Dell EMC is looking at providing software and services to both greenfield and experienced companies that have HPC but are looking to pivot capacity to dual-use traditional HPC plus deep learning. These companies may be looking to install applications that their employees can immediately put to use or use to build their own IP.
Dell Technologies has a key advantage in this space. The company’s servers are the de-facto answer for a large segment of the market which means it has an amazing amount of sales force reach along with install base data.
We were told that we can expect a bigger push in 2018 around the AI and deep learning offerings from Dell EMC. It was also hinted that the breadth of the infrastructure would include more than just NVIDIA GPUs in the data center and a variety of endpoint technologies at the edge. One example this AI push we saw with the Dell EMC PowerEdge C4140. That system had up to four NVIDIA Tesla V100 GPUs and dual Intel Xeon Scalable CPUs in a 1U form factor and was designed to address both deep learning and HPC workloads.
Final Words
This is a huge opportunity for Dell EMC if they are able to capitalize on their breadth of relationships. As AI models are trained and problems are solved, Dell EMC has the opportunity to leverage key insights and work across its massive install base. We are eagerly awaiting the announcements coming in Q1 2018 to hear more about the company’s key offerings in the space.
