Intel unveiled a new roadmap for its infrastructure processors for the network and the edge until 2026.
At its Intel Vision 2022 event in Dallas, Tech Radar Pro and other media have learned that the company plans to ship three new generations of infrastructure processing units (IPUs) over the next four years.
By the end of 2022, we can expect the previously announced Oak Springs Canyon FPGAs and Mount Evans ASICs to ship. This second generation range will be launched with a 200GbE network.
A little further, Intel plans to follow up with third generation IPUs (Hot Springs Canyon and Mount Morgan) with 400GbE, and a fourth generation with 800GbE, in line with the increasing demand for bandwidth.
Speaking to the press ahead of Vision, Patty Kummrow, CVP and GM of Network and Edge, outlined the opportunities opened up by IPUs, the first of which was spear by Intel last summer.
Basically, the idea is to help cloud and network providers free up CPU performance by offloading functions like storage and networking virtualization, which also offers various security benefits. Essentially, IPUs are almost identical to data processing units (DPUs).
“IPUs fundamentally change the way the data center can be architected; they’re a huge catalyst for the changes we’re seeing and the evolution of the data center,” Kummrow said.
“And we’re also seeing strong demand and applicability for these devices beyond hyperscale data centers, right out to the edge.”
With future generations of UIPs, Intel hopes to be able to help customers further optimize their data center resources and, by extension, increase their profitability.
The announcement of the IPU Roadmap follows the launch earlier this year of new Intel Xeon D processors for network and edge use cases.
The chips are built on the company’s Ice Lake platform and feature built-in AI and cryptography acceleration, built-in Ethernet, and various other features that address common network workloads and from the periphery.
At the time, Intel said Tech Radar Pro that the new chips deliver “breakthrough performance” in use cases such as security appliances, enterprise routers and switches, cloud storage, wireless networking, AI inference, and edge servers .