Leveraging flexible, multi Arm®-core i.MX 8 application processors, new embedded boards featuring chips like the i.MX 8M Plus boost the machine-learning abilities of edge devices.
How do i.MX 8 Processors Utilise Arm® Cortex® Cores?
NXP’s i.MX8 family of application processors comprises multi-core and heterogeneous devices. They are based on application processors from three Arm® Cortex®-A families: the ultra-high-efficiency Cortex-A35, high-efficiency Cortex-A53, and the high-performance Cortex-A72. You can choose from single or multi-core implementations and the devices also contain various focused accelerators and peripherals and a Cortex-M embedded core to handle low-level system monitoring and real-time tasks while the Cortex-A35 shuts off to save power.
The i.MX 8 family continues to grow, inspiring new Single-Board Computers (SBCs) from specialists such as CompuLab that bring exciting new powers to industrial automation and high-end consumer products.
What are the leading processors in the i.MX8 family?
When the i.MX8 family arrived in the market, we identified three key members that would drive significant advances in industrial and high-end consumer applications.
The i.MX 8X at the heart of the CompuLab CL-SOM-iMX8X system-on-module combines up to four ultra-high-efficiency Cortex-A35 cores and a Cortex-M4F with a GPU that accelerates graphic performance and the Tensilica® HiFi 4 audio DSP that allows keyword detection and speech recognition for hands-free interaction.
The i.MX8M in boards like CompuLab’s CL-SOM-iMX8 combines dual or quad Cortex-A53 high-efficiency cores with a powerful video processing unit (VPU). This can handle 4Kp60 video with H.264 and H.265 codecs as well as VP9, which is widely supported among browsers. High Dynamic Range (HDR) also makes this device ideal for medical imaging and inspection applications, and there are powerful features for streaming audio devices such as soundbars and networked speakers, two USB 3.0 interfaces, HDMI 2.0a, MIPI-CSI, and MIPI-DSI.
We also admire the i.MX8M Mini, with four Cortex-A53 cores, 400MHz Cortex-M4, video and graphics processors, and display and camera interfaces. It powers the CompuLab UCM-IMX8M, which measures just 28mm x 38mm x 4mm and consumes only 0.5-3.0W. With the processor’s MPCore TrustZone architecture, this board is an excellent choice for IoT applications such as smart home appliances. There is also support for separating memory domains, secure-region protection for on-chip RAM, and a Cryptographic Acceleration and Assurance Module (CAAM) that provides services such as content protection, integrity checking, and side-channel attack resistance.
What important new i.MX8 processors are available?
Now, the i.MX 8M Plus has arrived, combining the high-efficiency Cortex-A53 quad-core with the latest-generation Cortex-M7. With the new Neural Processing Unit (NPU), it’s the first in the series to feature a machine-learning accelerator. The NPU simplifies using neural networks for tasks such as multi-object surveillance, human-emotion detection, and pose recognition. Bringing AI to the edge, i.MX 8M Plus also integrates an Image Signal Processor (ISP) that can run image-correction algorithms on high-definition video in real-time.
Also new, the i.MX 8 QuadMax sits high in the performance spectrum. Packing two Cortex-A72s and four Cortex-A53s, as well as two Cortex-M4F embedded cores, it’s ideal for demanding machine-vision applications.
How do I choose the right i.MX8 device for my application?
The i.MX 8 family gives developers great flexibility to make every dollar count. The choice of single, dual, and quad Cortex-A cores lets you tailor the performance and cut power to the bone while the system remains ready to respond to any trigger thanks to the Cortex-M core. Often, outright performance is not the main concern and you may simply be looking for resilience; a system that can perform a set of focused tasks reliably for an extended period of time with minimal intervention.
One of the beauties of the i.MX 8 approach is the diversity of co-processors and accelerators that are available for handling graphics, video, voice, crypto, and – with the latest introductions – machine learning and artificial intelligence. You can get phenomenal performance and outstanding energy efficiency, in a solution that is cost-effective and compact.
When choosing any board for your project, keep in mind that software support holds the key to a great developer experience. After you have chosen your board, developing the software can consume as much as 70% of the project resources. Our boards come with fully documented BSPs, which cannot always be said for others, particularly those that are community-oriented and offer little or no support from the manufacturer.
At Anders we have close collaboration with the leading embedded board suppliers which ensures we get priority access to high-quality boards like the CompuLab i.MX8 portfolio.