AI Hardware
Running neural networks on constrained hardware: dataflow, number formats, and what each choice costs in logic and power.
An independent lab exploring AI hardware, programmable logic, and electronics.
Three areas, one bench: models, the logic that runs them, and the boards that carry the signals.
Running neural networks on constrained hardware: dataflow, number formats, and what each choice costs in logic and power.
RTL design, simulation, and timing closure on programmable logic, from a single counter to a full pipeline.
Schematic capture, board layout, power, bring-up, and measurement on the bench.
These are placeholder entries. Each card will link to a project page with schematics, build logs, and measured results.
Example entry. Replace with a real project: what it does, which board it runs on, and how far along it is.
Example entry. A board project shows the photo, the schematic, and what was measured at bring-up.
Example entry. A design study states the question, the approach, and the result, including what did not work.
Experiments, build logs, measurements, and lessons learned, written as the work happens.
Placeholder note. A build log records what was assembled, what was checked before power was applied, and what happened.
Placeholder note. A measurement states the setup, the probe, the numbers, and how far to trust them.
Placeholder note. An experiment states the question first, then the method, then the result.
Placeholder note. A lesson records a mistake plainly so it is not repeated.
NoSlack Labs explores AI hardware, programmable logic, and electronics. The work is hands-on: designs get built, put on the bench, and measured.
Projects and lab notes are published as they progress, including the parts that did not work.