NXP i.MX 9 processors to integrate Arm Ethos U-65 microNPU, EdgeLock secure enclave

NXP i.MX 9 Smart Cities

NXP i.MX 6 and i.MX 8 processors are widely used in industrial boards and systems-on-module, and the company has now teased a new family with i.MX 9 processors integrating Arm Ethos-U65 1 TOPS microNPU, as well as the company’s EdgeLock secure enclave for increased security. The company did not provide that many technical details, so we still don’t know which CPU cores, GPU, and exact peripherals will be found in the processor. But we do know the i.MX 9 processors will be manufactured with a 16/12nm FinFET class of process technology optimized for low power, and features the “Energy Flex” architecture that combines “heterogeneous domain processing (independent applications processor and real-time domains with a separate low-power multi-media domain), design techniques, and process technology to maximize performance efficiency”. That means most blocks of the processor can be turned off for low power audio or CAN networking use cases, and other industrial […]

It’s raining i.MX 8M Plus systems-on-module at Embedded World 2021

i.MX 8M Plus Systems-on-Module

NXP introduced i.MX 8M Plus AI SoC with a built-in 2.3 TOPS neural processing unit (NPU) last year, and we’ve already covered several early announcements about i.MX 8M Plus systems-on-module (SoM) with Variscite VAR-SOM-MX8M-PLUS and DART-MX8M-PLUS, TechNexion EDM-G-IMX8M-PLUS and AXON-E-IMX8M-PLUS respectively using SO-DIMM edge connectors and board-to-board connectors, as well as SolidRun i.MX 8M Plus SoM that was announced together with the HummindBoard Mate carrier board with dual Gigabit Ethernet. But as Embedded World 2021 Digital is taking place virtually until Friday, other companies have now made their own announcements of i.MX 8M Plus SoMs as the processor enters mass production this month, and since as far as I know, it’s pin-to-pin and software compatible with earlier i.MX 8M Nano/Mini SoCs, the update must have been easy. That means we’ve got a longish list of modules, and I have probably missed some. Supported operating systems are basically the same across […]

NVIDIA introduces lower cost Jetson TX2 NX SO-DIMM module

Jetson TX2 NX SO-DIMM Module

NVIDIA Jetson AI modules and developer kits range from the entry-level Jetson Nano module (5W, ~0.5 TOPS) to the high-end Jetson Xavier AGX module (30W, 32 TOPS). The higher-end modules usually come with a 400-pin board-to-board, while cheaper/cost-down modules like Jetson Nano and Jetson Xavier NX feature a 260-pin SO-DIMM connector and small form factor. But so far all Jetson TX2 modules came with a 400-pin connector, but this has changed with the introduction of the Jetson TX2 NX SO-DIMM system-on-module that is offered as a cost-down version of Jetson TX2 4GB module. NVIDIA Jetson TX2 NX specifications with highlights in bold or strikethrough showing the difference against Jetson TX2 4GB: Processor – Dual-Core NVIDIA Denver 2 64-Bit CPU and Quad-Core Arm Cortex-A57 MPCore processor @ up to 2.0 GHz GPU – 256-core NVIDIA Pascal GPU @ up to 1.3 GHz AI Performance – 1.33 TFLOPS Video Encode H.265 – […]

Lyra audio codec enables high-quality voice calls at 3 kbps bitrate

Lyra vs Opus vs Speex

We’re often writing about new video codecs like AV1 or H.266, and recently, we covered AVIF picture format that offers an improved quality/compression ratio against WebP and JPEG, but there’s also work done on audio codecs. Notably, we noted Opus 1.2 offered decent speech quality with a bitrate as low as 12 kbps when it was outed in 2017, the release of Opus 1.3 in 2019 improved the codec further with high-quality speech possible at just 9 kbps. But Google AI recently unveiled Lyra very low-bitrate codec for speech compression that achieves high speech quality with a bitrate as low as 3kbps. Before we go into the details of Lyra codec, Google compared a reference audio file encoded with Lyra at 3 kbps, Opus at 6 kbps (the minimum bitrate for Opus), and Speex at 3 kbps, and users reported Lyra to sound the best, and close to the original. […]

PoE powered AI Edge computer combines NXP i.MX 8M SoC with Google Edge TPU

AAEON has introduced several BOXER AI edge computers over the years either powered by an NVIDIA Jetson module, a HiSilicon HI3559A Arm SoC, or an Intel processor combined with Movidius Myriad AI accelerator. The latest model, BOXER-8521AI, combines an NXP i.MX 8M SoC with Google Edge TPU AI accelerator, and offers 1GB RAM, USB ports, a 40-pin external I/O connector, and 12V or PoE power inputs. AAEON BOXER-8521AI specifications: SoC – NXP i.MX 8M SoC quad-core Cortex-A53 processor with Cortex-M4F real-time core, 2D/3D GPUs AI Accelerator – Google Edge TPU ML accelerator coprocessor System Memory – 1GB LPDDR4x RAM Storage – 8GB eMMC flash, MicroSD card slot Video Output – 1x HDMI 2.0a port Audio – 1x Mic-in, 1x Line-out, optional 2x MEMS microphones Connectivity Gigabit Ethernet RJ45 port with PoE (802.3at) Optional? WiFi 2×2 MIMO (802.11b/g/n/ac 2.4/5GHz) & Bluetooth 4.1 plus two antennas USB – 2x USB3.2 Gen 1 […]

Vacos Cam AI Security Camera Review – Part 1: Specifications, Unboxing and Teardown

Vacos Cam teardown

As we’ve seen in our Reolink RLC-810A review, AI security cameras greatly reduce the number of false alerts generated by motion sensors, and the Reolink 4K security camera we tested was capable of people and vehicle detection. The Reolink model does not support WiFi, so I had to run a long Ethernet cable, plus a power extension to connect a 12V power adapter since I don’t own a PoE injector. But I’ve been given the opportunity to review another AI security camera, namely Vacos Cam, that also supports human detection, but relies on WiFi connectivity and can be powered by a solar panel removing the need for potentially long cable. As a bonus, it comes with eMMC for storage, so there’s no need for a MicroSD card. In the first part of the review, I’ll list the specifications, unbox the power panel and camera, and tear down Vacos Cam to […]

SolidRun CuBox-M 2-inch Arm Linux/Android mini PC packs a 2.3 TOPS AI accelerator

CuBox-M i.MX 8M Plus Mini PC

SolidRun first introduced the tiny CuBox Arm Linux mini PC in 2011, with the device based on a Marvell Armada 510 dual-core Armv7 processor, and followed in 2013 with the launch of the CuBox-i family powered by Freescale i.MX 6 single to quad-core Cortex-A9 processor. The company moved to NXP i.MX 8M processor in 2018 with the introduction of CuBox Pulse, and now they’ve introduced their first Cubox mini PC with a built-in AI accelerator. Meet CuBox-M mini PC powered by NXP i.MX 8M Plus quad-core Cortex-A53 processor with a 2.3 TOPS NPU. Cubox-M mini PC specifications: SolidRun i.MX 8M Plus System-on-Module (SoM) SoC – NXP i.MX 8M Plus Quad quad-core Arm Cortex-A53 processor @ 1.8 GHz with Arm Cortex-M7 up to 800MHz, Vivante GC7000UL 3G GPU (Vulkan, OpenGL ES 3.1, OpenCL 1.2), 2.3 TOPS NPU, 1080p60 H.264/H.265 video encoder, 1080p60 video decoder (H.265, H.264, VP9, VP8), Candence HiFi4 audio […]

Geniatech RK3568 Development Board to support Android 11 and Linux

RK3568 development board

RK3566 and RK3568 are the latest AIoT processors from Rockchip equipped with four Arm Cortex-A55 cores, a Mali-G52 GPU, and an impressive range of peripherals that make them suitable for all sorts of applications including AI edge computing and network video recorder (NVR). We already know Pine64 is working on Quartz64 SBC based on Rockchip RK3566, but I’ve just noticed Geniatech’s “RK3568 development board” that should offer even more features since the processor comes with extra I/Os compared to its little brother. Let’s have a look. Geniatech RK3568 development board specifications: SoC – Rockchip RK3568 quad-core Cortex-A55 processor up to 2.0 GHz with Arm Mali-G52 GPU supporting OpenGL ES 1.1/2.0/3.2, OpenCL 2.0, Vulkan 1.1, 0.8 TOPS NPU for AI acceleration System Memory – Up to 4 GB LPDDR4 Storage – Up to 128 GB eMMC 5.1 flash, 1x SATA 3.0 port, MicroSD card slot, M.2 socket for NVMe SSD (See […]

EDATEC Raspberry Pi 5 fanless case