@ terrence - Ok, at a high level anyway.
The processors that run boards like the Raspberry PI are not designed for Automation; they are designed for computation. This board's processor is designed for Automation. This is the first and main reason.
This module integrates the processor and external ram into a single package. This makes it harder for my competitors to reverse engineer my software by analyzing the signals between the cpu and the external memory. It was for this reason why I loved the single chip deigns of the STM32F4 class of devices.
The processor on this chip has 2 helper co-processors called Programmable Real Time Units (again because it's an Automation processor). When life and limb are on the line, you need to have a good level of determinism in your software. This is accomplished by executing that mission critical code on it's own 200mhz co-processor, that have direct access to the I/Os. This allows for pin toggling to take place in 1 cycle. They also have Direct Memory Access (DMA) to the system memory, so... there is that too.
The board itself is made by GHI, which has a strong tightly knit community behind it. Not like Raspberry PI which is big and spread out. Also, it uses castellated pins (and is flat on the bottom) as opposed to a SODIMM module. My application is subject to vibration. I can't risk using a system that could end up having intermittent connections
Cost: if he ever posts it. In order to turn the Raspberry PI into a system that would suit my needs, I would have to add stuff to it to make it do so. Depending on what GHI sets the price at, this module could have a lower total cost of ownership.