With EPYC Rome Chips, AMD Could Eclipse Intel in Datacenter -- Redmondmag.com

With EPYC Rome Chips, AMD Could Eclipse Intel in Datacenter

AMD's high-profile EPYC 7002 launch has datacenter analysts wondering if the end of Intel's long reign is nigh.

By John K. Waters
08/19/2019

Now that the hype has settled around the splashy San Francisco launch of AMD's EPYC 7002 "Rome" series CPUs, the salient impact of this release is becoming clear: The race is on for the datacenter, and Intel, the current market segment leader, is probably going to be eating AMD's dust.

In the circumspect language of a Wall Street analyst, the Rome chips are likely to have "significant industry reverberations that will play out favorably for AMD over the next several years." That cogent observation came from Hans Mosesmann, managing director at Rosenblatt Securities, in a client note following the early-August launch. But he also called the Rome launch "a history-changing event" that "may prove to be one of the biggest turning points in the history of Silicon Valley and computing," adding, "AMD's intentions are to own the datacenter. All of it."

"We're at the very, very early stages of a multiyear, market share-gaining dynamic, similar to what AMD enjoyed back in 2006, when they gained 25 percent of the enterprise market," Mosesmann told CNBC. "I think they have 1 percent on the market today. They'll probably get that 25 percent and then some over the next several years."

AMD's second-generation EPYC x86 chips are based on the company's 7nm process technology. They're a follow-on to the first-gen 14nm EPYC Naples CPUs, released in June 2017, and they were designed specifically for modern datacenter workloads.

The Rome processors feature up to 64 Zen 2 cores per SoC, deliver up to 23 percent more instructions per clock (IPCs) per core on server workloads, and up to four times more L3 cache compared with the previous generation, AMD said in a statem3ent. (Zen 2 is the micro-architecture at the foundation of AMD's chips.)

The processors are designed to deliver "record-setting" performance and lower total cost of ownership across a broad set of workloads, said Dr. Lisa Su, AMD's president and CEO, in a statement. "Adoption of our new leadership server processors is accelerating with multiple new enterprise, cloud and HPC [high-performance computing] customers choosing EPYC processors to meet their most demanding server computing needs."

During this month's launch event, held at San Francisco's Palace of Fine Arts, several customers and partners took the stage to talk about their planned Rome datacenter deployments, including Google, Microsoft, Hewlett Packard Enterprise (HPE), Lenovo, Dell, Cray and VMware.

Google's comments were typical: "AMD 2nd Gen EPYC processors will help us continue to do what we do best in our datacenters: innovate," said Bart Sano, Google's vice president of engineering. "Its scalable compute, memory and I/O performance will expand our ability to drive innovation forward in our infrastructure and will give Google Cloud customers the flexibility to choose the best VM [virtual machine] for their workloads."

Microsoft, Amazon Web Services (AWS), HPE, Dell and Twitter all announced plans to switch to the Rome chips or increase their current order count.

Intel, of course, isn't sitting on its hands while AMD marches into a market segment it has been dominating. Two days before the EPYC launch, Intel announced a "multiyear global collaboration" with Lenovo that will build on the two companies' longstanding partnership in the datacenter to "speed the convergence of HPC and AI."

"Intel is laser-focused on helping our customers spur innovation and discovery through the convergence of AI with HPC," said Navin Shenoy, executive vice president and general manager of Intel's Data Center Group. "Our extended collaboration with Lenovo combines the best of both companies' innovations to drive our customers' progress forward even faster."

However, as Mosesmann noted, that may not be enough. "From a market segmentation perspective, AMD's single-socket offerings will destroy (we had been using 'disrupt' in the past) Intel's 2-socket offerings," he said. "And interestingly, Intel's 4-socket solutions now may be irrelevant versus AMD's 2-socket offerings. EPYC 2 opens the floodgates to significant adoption that we suspect will eclipse the 24 percent share AMD achieved in servers back in 2006."

About the Author

John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [email protected].

Featured

Multi-Tenant Organization Hooks Released for Microsoft 365

Microsoft Entra ID's new multi-tenant organization capabilities have reached general availability (GA).
Real-World Ransomware Recovery

Here's how I was (mostly) successful in recovering a family PC infected with malware.
Microsoft Launches Small AI Models Family Phi-3

Recognizing that bigger isn't always better, Microsoft has released its smallest open AI models to date, Phi-3.
Microsoft and OpenAI Continue Global AI Expansions

Microsoft and OpenAI each continued their global expansion this month, with the announcements of new infrastructure investments and service rollouts in new regions, like Asia and the Middle East.
Cisco Warns of Brute Force Attacks Targeting VPNs and Firewalls

Cisco has revealed that a global increase in brute force attacks targeting Virtual Private Networks (VPNs) and other devices is currently taking place.