
Artificial intelligence, or AI, is a hot and trending topic right now in the world of technology and elsewhere. You may be familiar with popular forms of this technology recently, such as ChatGPT.
Now, technology companies like Intel are doing everything to ensure it can deliver the best AI experience on client devices so they don’t need to rely on the cloud for all their workloads.
This may also be Intel’s move to keep up with the tightening competition in AI, as experts believe that Intel is lagging in the race, as we previously reported here on PVP Live.
Intel has recently disclosed details that its upcoming Meteor Lake will feature a dedicated vision processing unit (VPU) to accelerate AI workloads. Here’s the news story.
VPU on Meteor Lake
Debuting later this year, Meteor Lake is a next-generation Intel CPU and the first to be based on the Intel 4 process while, at the same time, leveraging the company’s Foveros packaging technology in a chiplet-style design. This design paves the way for advanced power management, and each of the tiles can be targeted for particular workloads.
One of these chiplets is the CPU, developed to perform at its finest, and there is also a brand-new GPU that will support things such as ray tracing and XeSS in a low-power design.
However, the most significant focus for Intel right now is its new AI engine, the VPU, which is integrated into the processor. With this, Intel hopes it will accelerate client-side AI workloads, greatly reducing the computing prerequisites of AI inferencing.
Intel said a Meteor Lake system-on-a-chip (SoC) would integrate a CPU, GPU, and VPU, all bound to boost AI workloads. Firstly, the CPU helps quickly respond to light, single-inference, and low-latency AI. Then, the GPU is for AI in media and the 3D render pipeline.
What about the newly unveiled VPU? Well, VPU is a dedicated low-power AI engine that will be present in all Meteor Lake SKUs and meant to accommodate more intense AI tasks. It is a specialized processor designed to carry out specific AI workloads more efficiently than what a CPU or GPU can do.
Some of the tasks that VPU can handle include auto-framing, avatar recognition, background blur, gesture recognition, and dynamic noise suppression, as well as more advanced functionalities like Photoshop Neural Filters and real-time motion capture in Unreal Engine.
Demo
A demonstration of Intel’s VPU has also come out online, particularly that which was shared by YouTuber Hardwareluxx. It generally showed what the VPU could do.
Many who have watched the demo video were impressed by it, such as netizen Batuhan Cansın Özer who commented, “What a speed…”
How about you? What do you think of this news?