Manufacturing Matters Podcast: Ray Hoare on the Shift from FPGA Specialists to Systems Engineers
A camera that just moves pixels along isn't the point anymore. The point is the answer on the other end: is that a person or a stick in the road, is the laser weld pool the right size, is the part in spec? In this episode of the Manufacturing Matters podcast, Concurrent EDA founder Ray Hoare sits down with host Jimmy Carroll to talk about why 20 years into FPGA design, the job has quietly turned into something bigger — solving end-to-end systems problems, not just filling chips.
Ray walks through how AI-assisted development has compressed the prototyping cycle for custom machine vision and edge-compute algorithms, and where FPGAs still beat GPUs outright: sub-millisecond control loops, thousand-frame-per-second cameras, and applications that simply can't tolerate the roughly 40-millisecond latency of a typical embedded GPU.
The conversation covers real deployments across manufacturing, defense, and test and measurement: melt pool monitoring in additive manufacturing, laser tracking and 3D metrology, RFSoC for drone detection and defense applications, time-sensitive networking for multi-sensor synchronization, and the Allied Vision Alecs open smart camera, which lets engineers prototype an algorithm in Python with AI assistance before Concurrent EDA embeds it into a camera, frame grabber, or edge module.
Ray's advice for anyone with a high-speed data or machine vision problem they think is unsolvable: dream big, and tell him what your data source is, what you're trying to extract from it, and how fast you need the answer back.
Watch the full episode above or here → https://www.youtube.com/watch?v=UyhNgRUEzuQ.
Email