Getting started with HPC and Drones - Building an End-to-End System

[@Malik-1908]

Amazing! But I believe that optimization and taking advantage of all the SBCs features are not important anymore. Most companies these days just go with the highest standards SBCs and deploy them directly to the hardware. This will help us speed up the process of manufacturing and reduce the time and resources spent on R&D in hardware and software acceleration..

[@antshivrobotics]

Thank you for your reply! I think what you’re advocating makes a lot of sense, and you’re absolutely right-high optimization isn’t necessary for many use cases. A high-standard SBC (Single Board Computer) is more than sufficient for a wide range of applications. Optimization becomes critical mainly when you’re trying to speed up simulations, train AI models, or run compute-intensive tasks on edge solutions. As we move further into a future where AI plays an increasingly significant role, the cost and efficiency of running these algorithms on drones and other systems will depend heavily on squeezing every ounce of compute power from the hardware. This is where HPC concepts truly shine. In addition, focusing on high optimization often leads to better products in the long run or helps uncover new insights that are hard to grasp when relying solely on abstractions. I firmly believe in a yin-and-yang approach. Use abstractions and speed up development as much as possible, but also allocate some resources to R&D and peeling back those abstractions. This can lead to deeper understanding, new innovations, and benefits that pay off in the long term. Lastly, from a job market perspective, most top AI companies are prioritizing HPC and distributed computing skills in some form at almost every level. This demand is only going to grow. My hunch is that companies and individuals who invest in these skills and focus on high optimization for relevant applications will have a significant competitive edge in the long run. Just my two cents! 😊