Hi all – at an internship with Microsoft in summer 2019, I wrote a small set of tools for running simple static analyses of Relay programs and presenting their results in a human-readable format. I used it to gather and present data needed to make high-level architectural decisions about the design of a deep learning accelerator. It is not meant for the compiler’s static analyses. I recently revived it, as I need to do similar analyses for another accelerator design, and figured I’d also post it here, in case anyone else might find it useful 
1 Like