At the upcoming Computex event in Taipei, Intel demonstrated a working prototype of its Meteor Lake processor integrated with a general purpose processing unit (VPU).
Meteor Lake is Intel’s first chip to include a VPU, a dedicated AI engine that runs directly on the system-on-chip (SoC) to efficiently run AI models. Intel acquired his VPU technology in 2016 through its acquisition of California-based Movidius. It is accelerated inference integrated into a central processing unit and tuned for computer vision and deep learning applications.
With demand for this technology skyrocketing, Intel plans to ship millions of Meteor Lake units with dedicated AI engines over the next year. According to The Register, Intel vice president and client AI general manager John Rayfield said that for many PC workloads he needs a dedicated AI chip where AI exists. He added that currently the central processing unit (CPU) and graphics processing unit (GPU) do most of the heavy lifting.
In Meteor Lake processors, VPUs handle sustained AI and AI offload, while CPUs perform low-latency, simple inference tasks. GPUs, on the other hand, run jobs with performance parallelism and throughput. All other AI-related work is assigned to VPUs.
At the recently concluded Build event, Microsoft and Intel partnered to take advantage of the Meteor Lake processor’s “unique decoupled architecture” to enable AI-powered features for personal computer users. This includes new multimedia features such as automatic reframe in Adobe Premiere Pro, machine learning tasks, and scene edit detection.
“We are excited to work with Intel on the scale of AI that Meteor Lake brings to the Windows PC ecosystem,” said Pavan Davuluri, corporate vice president of Windows Silicon and Systems Integration at Microsoft.
