> That works well for typical digital signal processing applications, and not so well for generic number crunching libraries. Even then it breaks down with mor

> That works well for typical digital signal processing applications, and not so well for generic number crunching libraries. Even then it breaks down with more complex signal processing algorithms. FPGAs are great for simpler algorithms like FFTs, digital filtering, and motion compensation. They aren't quite as good at more complicated algorithms like edge detection. They really break down when you want to use ML-based algorithms.