Arm: See a Demo About Migrating a x86-Based App to ARM64
The hardware space is hotter than it’s been in ages, thanks to the disruptive influence of AI and its resource-gobbling workloads. Arm, founded more than 35 years ago — an eternity in tech — has long been supplying processor intellectual property (IP) to chip manufacturers, including Apple and Qualcomm.
In this On the Road edition of The New Stack Makers, Pranay Bakre, a principal solutions engineer at Arm, talked to Alex Williams, TNS founder and publisher, about how Arm is meeting the moment, and helping developers and their organizations migrate and run their applications on Arm-based technology.
Arm is poised for an eventful future. This conversation, recorded at KubeCon + CloudNativeCon North America, in Salt Lake City, took place before Monday’s report by Reuters, which claimed that the company is considering plans to make its own microchips.
But this episode of Makers focuses on what Bakre hears from customers about the challenges of migrating and running applications. It also includes a demo showing how to migrate an existing x86-based application to an ARM64 architecture, both in a cloud deployment scenario and in a containerized scenario.
AI and Hyperscalers
“We started this journey a few years ago, and now all the hyperscalers, like AWS, Google, Microsoft, have ARM-based instances. Oracle does too,” Bakre told Williams. For instance: Amazon Web Services’ Graviton is a processor based on Arm’s Neoverse IP, the company’s “server-class CPU,” as Bakre called it. Google’s Axion is also Arm-based.
“We are seeing a lot more deployments in this ecosystem, because we have built that ecosystem for years, and all the software ecosystem is present to support your migration journeys and your native application deployments to Arm.”
Arm’s architecture, Bakre said, “is power-efficient and cost-efficient by design” and that’s a major reason for its wide adoption. It’s also compatible with a lot of open source software, he added. In the Cloud Native Computing Foundation ecosystem, “90 to 95% of projects, of graduating, graduated and incubating projects, support native Arm binaries and support deploying their software” on Arm-based infrastructure.
At Arm’s booth at KubeCon, Bakre said, conference attendees frequently asked about Arm’s plans for the future, particularly in keeping up with the demands of running AI workloads.
“I’m sure that’s going to be tied to the hyperscalers when those instances come out,” Bakre said. “All of these cloud providers have … up to 60% performance benefits of a particular workload over either legacy architectures or all the previous-generation architecture. So those are the benefits people are seeing: I want to optimize my application more, and I want to make sure that it gets the best performance that is possible from that infrastructure.”
Check out the full episode to see the demo and learn more about Arm Neoverse and migrating applications to Arm-based infrastructure.
