Does a rewritten adder or multiplier still beat the reference once the ABC script is allowed to vary?

research/chip-ppa-script

Re-scores the chip-ppa-loop rewrites and sweep under five ABC scripts with every mapped netlist proved equivalent: the adder rewrite gain survives (0.80–0.86× the reference per script) but the multiplier's whole gain is the script (stylised, generic-gate).

5 tests

What it found

A follow-up to chip-abc-sweep and chip-ppa-loop: the earlier proposer files and a 16-adder and 8-multiplier sweep are scored under five ABC scripts, and every mapped netlist is proved equivalent to its reference (none failed). For the adder the best rewrite is 0.80–0.86× the reference under the same script, but the reference under its best script is already 0.61× its default score, so much of the headline gain was the script. For the multiplier no rewrite beats a * b under any script; the only gain (0.90×) is from the script. Stylised: two designs, one Yosys version, a session-written proposer, generic-gate counts rather than layout.

A figure from its own output

Figure produced by research/chip-ppa-script: ppa_scripts — chip-ppa-script/experiments/ppa_scripts.svg, written by the project's own experiment code.

Paper, code and results

More in LLM chip design loops

Does an equivalence-gated rewrite loop beat a parameter sweep on popcount, leading-zero count and absolute difference?Equivalence-gated rewrites of 16-bit popcount, 16-bit leading-zero count and 8-bit |a−b| against a no-model sweep: popcount 0.98×, |a−b| 0.86×, but on leading-zero count the sweep (0.56×) beats the session-written rewrites (0.80×) in 168 of 200 equal-budget draws (stylised, gate counts).
Does the rewrite-versus-sweep verdict for a 16-bit leading-zero count survive five ABC scripts?Re-scores chip-ppa-logic's popcount, leading-zero-count and |a−b| candidates under five ABC scripts with every mapped netlist proved equivalent: the clz16 sweep-versus-rewrite verdict flips with the script (best 0.35× the reference); stylised, generic gates.
Does fusing the accumulate into the partial-product tree beat multiply-then-add once the ABC script varies?A fused 4×4 multiply-accumulate: no written rewrite beats a*b+c under the default or three of five ABC scripts, but carry-save forms win by up to 9% cells×depth under dch and resyn2 (stylised, generic-gate).
Do the carry-save multiply-accumulate rewrites still win once mapped to real SKY130 cells, and does the generic-gate score predict that?The chip-ppa-mac carry-save MAC rewrites mapped to SKY130 HD cells: the plain a*b+c under its best script is the overall best, the generic-gate score ranks candidates only weakly like cell areaxdelay (Spearman 0.20; stylised, ABC delay estimate).
How rugged is the landscape of SKY130 area×delay over ABC pass sequences for a multiply-accumulate, and do simple searches beat random at 10–40 evaluations?An exhaustive 800-point table of SKY130 areaxdelay over ABC pass sequences for the MAC (optimum 11% below the best earlier script; needle-like landscape), with random search, hill climbing and session-written proposals compared at equal budget (stylised, preliminary).
How do MAC-lane count and bit width trade area against latency in a tiny quantised-MLP accelerator?Parameterised MAC-lane accelerator for an 8-8-4 quantised MLP, bit- and cycle-exact against NumPy in Icarus; lanes x bit-width sweep puts the cells-times-cycles minimum at 4 lanes (stylised, generic-gate counts).

All 40 LLM chip design loops projects →

Other research topics

Proper scoring & elicitation28 Effort & peer prediction9 Verification games & slashing19 Detection & tolerances14 Markets as routers13 Prediction-market design9 Paying contributors8 Decentralised training systems23 Digital twins & sim-to-real164 LLM chip design loops40 Robot manufacturing loops53 Ambient AI assistant46 Self-replicating manufacturing34 Simulation-first design64 Engineered clothing45 Task-specific AI accelerators40 Algorithmic chip-design loops42 Open-source purpose-built robots44 The software-engineering loop43 Adversarial human imitation41 Open-source residential construction41 Autonomous multi-angle capture34 Sensing, imaging and 3D31 Energy supply and dense energy storage39 Robot societies among people31 Agentic co-design of low-cost robots27 Laundry-hanging robots25 Machine learning on qualitative data23 Robots on the building site27 Civilizations & group psychology1 History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.