Linux on the aeMB

Okay, it’s been a while since I stopped writing temporarily. I think that I the short hiatus from blogging was a good one. I can now return to writing about totally random and impersonal stuff like I used to. So, I think that I will kick things off by talking about recent developments on the aeMB processor.

A French company had decided to take the aeMB and build a modular routing platform around it. The platform is intended to work as a network router that can be used for mesh Wi-Fi networks, as an alternative to ISPs home gateways, and as a development platform for SoC projects. They have not reported any bugs and have instead been reporting a string of successes with the aeMB.

Firstly, on the hardware front. As a result of using my processor, they were able to reduce their total resource consumption. They previously used a Microblaze based platform that consumed about 67% of the total FPGA resources but by building an equivalent platform around the aeMB, they ended up only using about 40% of the FPGA resources. This is a very significant savings which gives them plenty of room for expansion.

In fact the US company that chose to use the aeMB at the heart of their platform, also chose it because of its small size and performance. This was done after comparing it with offerings from other people. The small size comes from the fact that care was taken while designing it to differentiate the aeMB from other embedded processors by being small and efficient instead of being a raw number cruncher.

Next, on the software front. The French company had previously used Linux on their platform and they have reported success with running Linux on the aeMB. Put differently, the aeMB is capable of booting Linux, today! Although I have always been confident that the aeMB can run Linux, I have never actually tried it before as none of the other users have needed to use Linux in their applications.

This has always been a sticking point at my side and has always stayed on my long TODO list. However, this French company has just saved me a tonne of time by actually proving that Linux works on it. They are planning to run OpenWRT on the aeMB, which is a Linux distribution originally made for Linksys routers but is now available on a variety of platforms.

Finally, on the enhancement front, everyone wants something different. This means that the TODO list is just getting longer and longer. The US people want to have a faster algorithm performance while the French people want to have better memory protection and some other people have asked for hardware debugging support.

For the moment, I don’t really have the time to actually work on any of the enhancements. So, I will just continue to let them use it as is, and work out any existing bugs. Then, at some point in the future, I will definitely find ways to add in all those things. At that time, it would become a full featured high performance embedded processor.

Both the US and French companies would like to see their enhancements built in and have offered to help in whatever way they can. I think that I will take them up on their offers at some point in the future. I just want to grow the aeMB thoughtfully and not jump into any random decisions on where to take it. Maybe it’s time that I published a road map and solicit some feedback on it.