‘AI becoming the fastest growing workload in data centres’

In line with the global launch of the second generation of Xeon processor in San Francisco. Chip maker Intel has introduced one of the widest range of innovations in its data-centric portfolio in the history of Intel Corp.

The legacy of Xeon has been existing over the last decades for Intel, however, after a gap of four years, the company has launched its 2nd-generation Intel Xeon Scalable processor. It claims to have made significant improvements over first generation processor, along with computing, memory and networking slew of products.

Speaking on the sidelines of the product launch in Bengaluru, Prakash Mallya, Vice President, and Managing Director, Sales and Marketing Group, Intel India, explains how the new offerings will boost up the computing performance that will pave way for emerging technologies like AI, ML, 5G, and cloud. Some edited excerpts…

Intel has refreshed its data centre offerings in tune to the digital transformation. Can you tell us what were the broad trends that pushed Intel to launch these products?
My association with Intel is over the last 19 years old and we have never seen this width and breadth of portfolio until today. From the India point of view, globally by next year, we will have every human being on earth will create 1.7 megabytes of data every second and 200 billion connected devices. Seeing the proliferation of data, the data center needs to be highly scalable and agile to meet the growing demand for data.

Moreover, with the democratisation of AI in the last few years, a lot of contribution has been made by large data-led companies. At Intel, our endeavour is to increase the benefits of AI to large as well as small companies. Another driving force is the rollout of 5G. It will create an impact on the intelligent edge, it will force operators to create a network which is agile, scalable and has more computing power and performance. Intelligent Edge is critical for 5G.

The last trend is cloud, it is the most matured one, out of the three. Today, cloud has transformed to enterprise, to network, to edge truly becoming widely accepted service delivery model. When you look at these trends and see the opportunities ahead from a data centre point of view, you have to re-imagine the entire infrastructure. It helps our organisation to move data faster, store more data and process it quickly. From data movement standpoint, network traffic is accelerating within a rack and across the data centre.

Please shed light on the major enhancement Intel has made to second generation Xeon processor? Is there a change in its positioning and business model?
The biggest enhancement from the first generation to the second generation is making of specific workloads optimised for our platform. For instance, the second generation Xeon, with deep learning, boosts performance 14x. It will cater to the growing need of AI, since AI is the fastest growing workload in the data centre. In network functions, the transformation of networks is happening from proprietary to open-source based on x86.

Through optimisation, we have been able to make specific workloads grow faster. Storage innovations in the second generation Xeon platforms, along with Intel Optane DC persistent memory, is game-changing in terms of big data analytics, in-memory computing in real time for management. Any of use cases requires high capacity and high performing memory. Until now, it was not feasible economically to deploy. The DC memory is estimated to be 50 per cent of the cost of DRAM. Lastly, we are exposing our customers and partners to bring out more innovations in products and solutions. Going by the adoption of the first generation Xeon, we are confident to see more appetite from our partners and customers in India. The market needs will push the ball forward in that direction, as customers and partners want to innovate and disrupt themselves as well as the industry.

How do you see the evolution of AI in India, with the introduction of Intel’s 2nd-Generation Xeon Scalable processor?
AI is not an application, it will be embedded in the applications. AI inferences get embedded in the application which will accelerate the optimisation with Intel DL Boost technology. It impacts the performance significantly and makes it flexible to deploy as well. We would see deployment become faster; our partners and customers can have more AI driven opportunities. Overall I see only two per cent of created data is being analysed as it requires more compute performance. Now, AI inference workloads like image-recognition, object-detection, and image-segmentation within a data centre, enterprise, and intelligent-edge computing environment. Intel has worked extensively with ecosystem partners to optimise frameworks and applications that take full advantage of Intel DL Boost technology. Customers can choose enhanced tools like OpenVINO to ease deployment.