color me skeptical on whether nuclear physics systems need software triggers. At the lhc I see the benefit to broadcasting complex data and event types across

color me skeptical on whether nuclear physics systems need software triggers. At the lhc I see the benefit to broadcasting complex data and event types across many instruments. It’s my opinion that misunderstandings of captured data lead to serious science errors and I can see why ppl want to merge into software architectures some of the front end work. Imo the best thing is that one will be able to hand this over to deep learning pipelines in an easier way, however latency becomes a real issue when flowing this much data. I suspect one will return to dense fpgas for efficiency and scalability. These signal problems have true information limits and sample rates / multichannel acq will force event driven pipelines or a new hybrid architecture. Source: I do front-end detector computing in both hardware and software front-ends professionally. Event driven paradigms are vastly superior for cutting edge high resolution and multichannel systems, “software” front ends are good for inspection but streaming data may be less useful long term. Bespoke systems might be software oriented but not for more than a few deployments, or not without substantially wide applications.