Methodology

Our methods depend on the question. A tool’s documentation can establish what it says it accepts and produces. A recorded run can establish what one version did with one input. A repository at a pinned commit can establish what files it contains. None of those, alone, proves the others.

Runnable examples

Each how-to example is one original SystemVerilog module with its testbench, or one script, printed in full. It is compiled and executed as printed in named open-source tools, usually Icarus Verilog and Verilator, with commands, versions, return codes and literal output recorded and published with the page’s data. The printed source is hashed so the run can be tied to exactly those bytes. A PASS line means the testbench’s own checks held in that simulator; the page states what the checks do not cover.

Simulator observations

The simulator fixtures are small original programs that each isolate one language behaviour: signedness and width, unknown values, initialization, assignment ordering and seeded randomness. Identical source bytes are run in each pinned tool with recorded options, and only literal output lines are compared. Each tool is also run twice to confirm it repeats its own result. Differences are explained against the configuration and the language standard; they are never scored as defects.

Tool and IP records

The EDA handoff and open-IP records are documentation reviews. Each field comes from the tool’s own documentation or the project’s repository at a pinned commit, with the source and date in the ledger. A field the sources do not establish is marked unverified, which means we did not confirm it, not that it is absent. No listed core was synthesized or executed, and no end-to-end flow was run through the listed tools.

The IP finder

The finder filters a static table in your browser. It makes no network requests while filtering, accepts no uploads and stores nothing; repository links open the source project. Its behaviour was checked in a real browser and the observation is published with the data.

What we publish

Compilations and observations are CC BY 4.0, and original code is MIT licensed. Each asset’s METHOD file, hash list and source ledger are published with it. Those licences do not relicense third-party tools, cores, standards or documentation.