Intel unveils AI-focused Movidius VPU chip
Intel on Monday announced its next-generation Movidius vision processing unit with improved processing capabilities for edge devices such as drones, VR headsets, smart cameras, wearables, and robots.
Movidius, an Intel subsidiary acquired by the chip giant back in September, develops sight capabilities for machines and PCs. Its latest VPU is the Myriad X system-on-chip that's equipped with a dedicated Neural Compute Engine to support deep learning inferences at the edge.
The on-chip hardware block is designed specifically to run deep neural networks at high speed and low power. Intel says the deep neural network accelerator can achieve over 1 trillion operations per second of computing on DNN inferences.
"We're on the cusp of computer vision and deep learning becoming standard requirements for the billions of devices surrounding us every day," said Intel VP Remi El-Ouazzane. "Enabling devices with humanlike visual intelligence represents the next leap forward in computing."
Intel touts the Myriad X as ideal VPU for autonomous device platforms due its tiny form factor and on-board processing power. In addition to its Neural Compute Engine, Myriad X also comes with 128-bit VLIW vector processors, increased configurable MIPI lanes, enhanced vision accelerators, and 2.5 MB of homogenous on-chip memory.
Intel said the Myriad X VPU comes with an SDK that includes a neural network compiler and "a specialized FLIC framework with a plug-in approach to developing application pipelines."
The Movidius Myriad 2 Vision Processing Unit will be used to run deep neural networks for higher accuracy, local video analytics.
The Movidius Neural Compute Stick compiles, tunes and accelerates neural networks at the edge.
The deal could push Intel further into next-generation technologies including VR, drones and artificial intelligence.