About
Above the RTL · Notes is the technical reference site for Above the RTL, a Substack publication on AI and chip design written from inside the industry.
The Substack carries the narrative — what's happening in AI-for-silicon, what works, what doesn't, what's actually worth an engineer's attention. This site carries the reference material that sits behind those posts: the physics, the MTBF arithmetic, the archetype catalogs, the long-form work that would bog down a post but matters to engineers actually closing sign-off.
Current focus
The initial library is building out around clock domain crossing — synchronizer physics, the archetypes of CDC failure, MTBF analysis across frequency regimes, methodology for putting AI into a CDC flow, and the correct-by-construction vs. safe-under-assumptions question that every reusable block eventually faces. Each reference note here is a companion to a narrative post on the Substack.
Adjacent topics that will land in the library as the series grows: formal verification methodology, assertion-based design, Design by Contract for hardware, and evaluation of AI models on chip-design tasks.
How this site relates to the Substack
Posts on abovethertl.com cover the argument. Notes here cover the analysis. Where a post says "the full derivation is in a companion note," that note lives on this site. Readers who want the narrative can stay on Substack; readers who want the physics follow the links here.