On-Device AI and Computer Vision

Myriad™ X: Ultimate Performance at Ultra-Low Power

Myriad™X is the third generation and most advanced VPU from Movidius™, an Intel company. Myriad™ X is the first VPU to feature the Neural Compute Engine - a dedicated hardware accelerator for deep neural network inferences. The Neural Compute Engine in conjunction with the 16 powerful SHAVE cores and an ultra-high throughput intelligent memory fabric makes Myriad X the industry leader for on-device deep neural networks and computer vision applications. Myriad™ X has received additional upgrades to imaging and vision engines including additional programmable SHAVE cores, upgraded and expanded vision accelerators, and a new native 4K ISP pipeline with support for up to 8 HD sensors connecting directly to the VPU.

As with Myriad™ 2, the Myriad™ X VPU is programmable via the Myriad Development Kit (MDK) which includes all necessary development tools, frameworks and APIs to implement custom vision, imaging and deep neural network workloads on the chip.

Key Benefits

Dedicated Neural Compute Engine
16 High Performance SHAVE Cores
Enhanced ISP with 4K support
New Vision Accelerators including Stereo Depth
View more details on chip

Neural Compute Engine

Myriad™ X is the first VPU to feature the Neural Compute Engine - a dedicated hardware accelerator for running on-device deep neural network applications. Interfacing directly with other key components via the intelligent memory fabric, the Neural Compute Engine is able to deliver industry leading performance per Watt without encountering common data flow bottlenecks encountered by other architectures.

16 Programmable SHAVE Cores

Myriad™ X features 4 more programmable SHAVE cores compared to Myriad™ 2. This additional 30% of programmable compute can be used to run traditional computer vision workloads, or can complement the Neural Compute Engine by running custom layer types for CNN applications thanks extensive support for sparse data-structures.

Enhanced Vision Accelerator Suite

Movidius has added a new suite of vision accelerators to Myriad™ X, including a new stereo depth block that is capable of processing dual 720p feeds at up to 180Hz. With the suite of vision accelerators available, it is possible to offload key vision workloads onto fixed-function hardware, resulting in higher performance at lower power.

Flexible Image Processing and Encode

Myriad™ X features a fully tune-able ISP pipeline for the most demanding image and video applications. Myriad™ X also features hardware based encode for up to 4K video resolution, meaning the VPU is a single-chip solution for all imaging, computer vision and CNN workloads.

Support for Multiple VPU Configuration

New PCI-e interface allows device makers to implement multiple Myriad™ X VPUs in a single device, delivering even more performance flexibility.

Neural Compute Engine: Hardware Based Acceleration for Deep Neural Networks

Myriad™X features the all-new Neural Compute Engine - a purpose-built hardware accelerator designed to dramatically increase performance of deep neural networks without compromising the low power characteristics of the Myriad VPU product line. Featuring an array of MAC blocks and directly interfacing with the intelligent memory fabric, the Neural Compute Engine is able to rapidly perform the calculations necessary for deep inference without hitting the so-called "data wall" bottleneck encountered by other processor designs. Combining the neural network performance of the 16 proprietary SHAVE cores with the neural compute engine, Myriad X delivers 10X the performance compared to previous generations*.

Key Benefits

Native FP16 and fixed point 8 bit support
Rapidly port and deploy neural networks in Caffe and Tensorflow formats
End-to-End acceleration for many common deep neural networks
Industry-leading Inferences/S/Watt performance

Related Videos