Feature Description
Last updated
Was this helpful?
Last updated
Was this helpful?
Auvidea ships some development systems with a pre-configured SSD. This contains a bootable partition which can be larger than the storage capacity of the integrated eMMC flash memory on the compute module. This allows the installation of all tools and supporting applications in the boot partition:
Linux
Jetpack 4.x
A large selection of SDKs like listed in appendix D (CUDA, VisionWorks, DeepStream and more)
3rd party applications and demo programs
Sample video files to immediately try out the AI models
The SSD typically included is a 128 GB NVMe SSD by Transcend (model: TS128GMTE110S). It offers very high performance as it connects to the compute module with 4 PCIe Gen2 lanes (5 Gbit/s).
Full length 2280 SSDs must be inserted carefully as the SO-DIMM socket of the compute module blocks tilting. Please slide the SSD in carefully from the side and ensure, that it is seated properly before applying power. Secure the SSD with a M3 (nylon) screw. Metal screws must be treated with care, so that the head of the screw is small enough that it does not touch any traces or components on the compute module.
High power SSDs may require extra cooling by installing a thermal pad between the SSD and the compute module or the carrier board.
The crypto chip has been added to support software licensing and copy protection. Please check the data sheet of the manufacturer for details.
Model: ATSHA204A-MAHCZ-T
An MCU for additional power management and watchdog functions is optional. It is standard on some extended versions of the JNX30. The MCU can provide features like:
Powering control (up and down) of the compute module
Brown out protection
Clean boot after short power interruptions
Watchdog support: it can connect to the debug console of the compute module
Remote power cycle via LTE (requires LTE module with UART interface – like SIMCOM modules)