reComputer J4012 is a mini PC or “Edge AI computer” based on the new NVIDIA Jetson Orin NX, a cost-down version of the Jetson AGX Orin, delivering up to 100 TOPS modern AI performance.
The mini PC is based on the Jetson Orin NX 16GB, comes with a 128GB M.2 SSD preloaded with the NVIDIA JetPack SDK and offers Gigabit Ethernet, four USB 3.2 ports, and HDMI 2.1 output. Wireless connectivity could be added through the system’s M.2 Key E socket.
reComputer J4012 / J401 specifications:
- SoM – NVIDIA Jetson Orin NX 16GB with
- CPU – 8x Arm Cortex-A78AE core @ up to 2.0 GHz with 2MB L2 + 4MB L3 cache
- GPU/AI
- 1024-core NVIDIA Ampere GPU with 32 Tensor Cores @ up to 918 MHz
- 2x NVDLA v2.0 @ 614 MHz
- PVA v2 vision accelerator
- 100 TOPS AI performance (sparse)
- Video Encoder (H.265)
- 1x 4Kp60 | 3x 4Kp30
- 6x 1080p60 | 12x 1080p30
- Video Decoder (H.265)
- 8Kp30 | 2x 4Kp60 | 4x 4Kp30
- 9x 1080p60 | 18x 1080p30
- System Memory – 16GB 128-bit LPDDR5 102.4GB/s
- Storage – 128GB NVMe SSD via M.2 Key M socket
- Video Output – HDMI 2.1 up to 8K resolutions
- Camera I/F – 2x 15-pin 2-lane MIPI CSI connectors
- Networking – 1x Gigabit Ethernet RJ45 port
- USB – 4x USB 3.2 Type-A (10Gbps) ports, 1x USB2.0 Type-C device port
- Expansion
- M.2 Key M socket (for SSD)
- M.2 Key E socket (for wireless connectivity)
- 40-pin expansion header
- 12-pin control and UART header
- CAN Bus
- Misc – 4-pin fan connector (5V PWM), 2-pin RTC socket + RTC battery holder
- Power Supply
- 9-19V DC via power barrel jack; 12V/5A power adapter
- 10W and 25W power modes
- Dimensions – 130 x 120 x 58.5mm (Aluminum case)
- Temperature Range – -10°C to 60°C
Some of the connectors listed in the specifications won’t accessible with the reComputer J4012 due to the enclosure. The mini PC and its J401 carrier board are an evolution of the reComputer J101/J202 carrier boards for Jetson Nano/NX/TX2 NX SoM found in the reComputer J2021/J2022 Edge AI computers. Seeed Studio also launched the reComputer J4011 with the same specifications except for the NVIDIA Jetson Orin NX 8GB module with just six Cortex-A78E cores, 8GB RAM, a single NVDLA accelerator, a lower GPU clock, and a maximum of 70 TOPS of AI inference performance.
The system comes pre-installed with NVIDIA JetPack 5.1 based on Ubuntu 20.04, and supports NVIDIA tools such as the DeepStream SDK, the TAO toolkit, and the Riva SDK for Speech AI applications, as well as third-party solutions such as Edge Impulse, RoboFlow, and AlwaysAI. Seeed Studio has published a wiki with the resources to help developers get started with the system.
The reComputer J4012/4011 mini PC can handle image recognition, object detection, pose estimation, semantic segmentation, video processing, and more making it suitable for applications such as traffic management, Industry 4.0, Smart Retail, Robotics, Smart Agriculture, and medical imaging analysis. Seeed Studio is taking pre-orders for the reComputer J4012 for $899.00 with shipping scheduled to start around mid-march. The lower-end reComputer J4011 is not for sale at this time, probably because the NVIDIA Jetson Orin NX 8GB is not available just yet.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
This has got me wondering if it would be better to get a different x86 or ARM mini PC and add an AI accelerator yourself, in an M.2 slot. I think the best you can do is Hailo-8 with 26 TOPS for around $200.
Also, AMD’s Phoenix Point APUs will include an “XDNA AI Engine”, potentially capable of around 18-19 TOPS (based on AMD’s estimate of it being 20% faster than Apple M2’s 16-core 15.8 TOPS Neural Engine).
The Axelera M.2 card that should be available soon is said to deliver performance similar to the Jetson Orin NX 16GB used here.
https://www.cnx-software.com/2023/01/02/150-axelera-m2-ai-accelerator-214-tops/