Today SolidRun announced their latest embedded System on Module (SoM) board which aims to "dethrone" the Raspberry Pi as the go-to hardware solution for development and prototyping. With a design similar to the recent Raspberry Pi Compute Module 4 the i.MX8M Plus comes in a dual or quad core configuration and with compatibility for existing models and software resources. The big selling point of this board is an onboard Neural Processor which will provide much more computational power when used in TensorFlow machine learning projects. The Raspberry Pi 4 can do at best 9 FPS in a TensorFlow Lite unless used with a Tensor Processing Unit such as those from Google's Coral project.
The i.MX8M requires carrier boards with which some of their onboard functionality is broken out for easy development. The three new carrier boards the HummingBoard-M Pulse, Ripple and Mate provide dual Gigabit Ethernet (Mate and Pulse) as well as an additional MIPI-CSI camera connector for stereo vision projects. Carrier boards are necessary for breaking out and developing projects based on the i.MX8M Plus CoM. Once we learn how they work we are free to develop our own carrier boards to break out the functionality that we require.
With prices starting from $75 for a quad core model with 1GB of RAM and 8GB eMMC but no wireless connectivity or carrier board we must look to the $150 model with Wi-Fi and Bluetooth as well as a carrier board.
