What rigpapa said.
I’ve spent 45 years designing embedded control systems for entertainment and architectural lighting. My first system in 1975 had a DEC PDP-8/A minicomputer with 16K words (12-bit word) of magnetic core memory. The entire operating program was 8K words, with the remaining 8K for cue storage and executed around 250K instructions per second. It ran a major Broadway musical for 13 years with no problems.
I did a small controller in 2005 (32-bit ARM, 180 Mhz, 128MB RAM, 16MB FLASH, our own embedded Linux) that takes in steaming Ethernet at 100 Mb/s and output two 250KB/s RS-485 streams while doing substantial processing on the data. The system never got above 0.1 utilization, and my lab unit has been on the bench without rebooting since 2006. It can be done.
And a big contributor is also constant external and internal pressure for new features, which I'm sure Vera's engineering team gets in volume, never leaving the time to go back and really review and clean house, unless something is on fire in that area. There's no reward for going back and making it work better; just make it work and move on, turn and burn. That is what I think Melih should change, because it's affecting more than just performance.
Bingo! And this is endemic in the software industry—it’s not just a Vera/Mios problem. I wish Melih and his new combined team all the best, but we need to be realistic. Any merger is a huge challenge, and it will take time to build the new organization