AMD, the Santa Clara, California-based semiconductor major, has been present in India for almost 15 years now.
AMD, the Santa Clara, California-based semiconductor major, has been present in India for almost 15 years now. It has a strong R&D presence, coupled with a strong engineering base here. “We are not only focused on commercial customer demands with our technology on the client side, but also on supporting government set up the supercomputing infrastructure,” Mark Papermaster, CTO and senior vice-president, Technology and Engineering, AMD, tells Sudhir Chowdhary in a recent interaction. Excerpts:
How has AMD reinvented itself to re-ignite competition in the CPU, GPU and server markets?
More than five years ago, we set out to return AMD as a top competitor in high performance. We had all the knowhow, but needed focus on the right product goals, and our engineering processes. With the new Zen x86 CPU and Vega GPU, we introduced high-performance compute and graphics architecture. We have invested heavily in our software libraries to enable customers and users to get to the most out of our products. We introduced Infinity Fabric, which is essentially the plumbing at the chip level to make it all work together seamlessly. Ryzen is a showcase example with its award-winning CPU and GPU performance, inclusive feature set, and energy efficiency—it excels at everything from gaming to demanding compute workloads.
Intel’s CEO confirmed recently that the company would lose at 15-20% market share in datacentre to AMD’s EPYC chips by the end of this year. How has AMD managed to achieve this historic increase when it only had 1% market share in 2016?
The beauty of EPYC, our x86 server CPU, is that it seamlessly runs the plethora of x86 server applications while bringing in excellent advantages to the customer for more efficient computing. To date, there are more than 50 EPYC based platforms in the market, and we are actively working with OEMs, system integrators and channel partners to expand growth. Given the design wins we have today, we are on our way to mid-single digit unit share exiting the year and see a path to double digits going forward.
Can you elaborate on AMD’s Zen architecture, which is scalable from the company’s CPU chips to datacentre? How will this bring multi-platform leadership to users?
Zen delivers a 52% performance increase over the previous generation core, which is the biggest leap in the company’s history. Zen combines high-performance data throughput, instruction execution and memory latency for optimum compute efficiency. The CPU is built in the leading edge FinFET semiconductor process, which is a key enabler of scale along with AMD’s design methodology. Zen is deployed across all of AMD’s CPU business, from enthusiast desktop platforms to enterprise-class servers and notebooks, to embedded and semi-custom products.
Since you are the only company to have both GPU and CPU capabilities, how can you combine them to drive innovation in supercomputing?
High performance computing and graphics technologies are, once again, playing a defining role in major technology trends that are reshaping the world around us. AMD is at the heart of this technological revolution as the only company in the world that can design both. EPYC is designed to efficiently pair with multiple GPUs for the most demanding machine learning, training and inference workloads with no additional chips required. The key to fully unleashing this capability begins with software enablement and we’ve brought competition to Machine Learning with our Radeon Open Ecosystem (ROCm). Additionally, we introduced the AMD Ryzen Embedded V1000 in February this year. It combines the computing power of Zen with Radeon graphics on a single integrated power efficient chip, enabling performance advantage at edge of network and embedded apps.
India is waking up to the benefits of supercomputing and has invested in the National Supercomputing Mission (NSM). How can AMD play a key role here?
As the only company in the world that can provide both high performance CPU and GPU, AMD is uniquely positioned in the HPC market place which has moved from significantly using CPUs to a hybrid CPU + GPU environment being driven by needs of highly parallelised compute requirements driving the HPC usage. Our strong R&D presence, coupled with a strong engineering base in India, provides partnership opportunities for the government to not only embark upon the National Supercomputing Mission, but also to execute it successfully.
How important is India to your global strategy?
With a large pool of trained manpower, a cost environment which fosters growth and a favourable government policy – India is the pivot to global hi-tech market. We recognise India’s growing influence not just as a consumer but also as a producer in the high-tech arena and we remain committed to the long-term aspects of being in India. We set up our first development centre in Bengaluru in 2004 and Hyderabad development centre in 2006. Both centres today play key roles in our global product development. On the business side, with AMD technology catering to all the compute segments, we are well-positioned to gain significant market share in India. We are already doing well in the commercial client segment and soon AMD server products will be here, further providing impetus to our market presence. In the areas of Machine Intelligence and Deep Learning, India is uniquely poised to leverage this opportunity with not just a base to develop these applications, but also as a user of these technologies to overcome development challenges.