[07]Journal· 4 entries
Notes from the work.
Engineering write-ups, design decisions, and the occasional thing I got wrong. Search runs full-text across titles, summaries and body copy.
- 05 Oct 2026Engineering
What an LDPC bit-flipping decoder actually spends its energy on
Throughput numbers get published; energy per decoded bit rarely does. Here is what changed when we stopped optimising iterations and started optimising data movement.
- ldpc
- ecc
- nand-flash
2 min read - 19 Sept 2026Engineering
Why the first draft of an ASIC spec is an algorithm, not a block diagram
Block diagrams drawn before the algorithm is settled are fiction. A note on the order I have settled into after two tapeouts.
- asic
- design-methodology
- rtl
1 min read - 02 Sept 2026Notes
Reading a paper properly: three passes and a question
The two-pass method I abandoned. Three passes, one question written down before you start, and a much better hit rate on papers worth your time.
- reading
- methodology
- writing
1 min read - 11 Aug 2026Engineering
Optoelectronic decoding, in one page
A compressed explanation of the JETCAS paper for anyone who knows what an error-correcting decoder does but not what an optical one is.
- optoelectronics
- compute-in-memory
- ecc
1 min read