Research
Projects, publications, open questions, and why I do research.
Projects & publications
- macro_gen: sizing standard cells by driving ngspice. Ongoing. Try it in the lab.
- Hermes: generating on-chip networks in Bluespec from a spec. Ongoing.
- PARISCV: profiling RISC-V applications to suggest custom instructions. OSCAR’24.
- AntGen: generating patch antennas in HFSS. IEEE MAPCON’23.
Open Questions
The problems I keep returning to - still discovering, still forming.
- Can physical constraints (placement, wire delay, congestion) be expressed as first-class parameters inside a hardware compiler, rather than deferred to back-end engineers?
- Where is the right boundary between programmable processor and fixed accelerator - and can that boundary be determined algorithmically from application behavior?
- Can we build scalable tools and methodologies that help us perform design-space exploration?
Why I do research
The longer version, on curiosity, courage and abstraction, is its own essay: On research →