VLSI Korea

DESIGN LAB / 04 · OPEN PPA ATLAS

SYNTHETIC DEMO

Turn design intuition
into a verifiable map.

Read performance, area, and power together and trace the generation context behind every point. Every design name and value currently shown is synthetic demonstration data—not an observed benchmark result.

VISIBLE RUNS
08
all synthetic
DESIGN FAMILIES
04
fictional RTL
PUBLICATION RULE
PROOF
or it stays private
Public synthetic dataset · not measured or tapeout results

To prevent confusion with observed results, every value—including design, PDK, commit, and runtime—is explicitly labeled fictional or synthetic.

ATLAS / 01

PPA design space

Performance increases to the right and cost increases upward. Highlighted points are synthetic Pareto candidates not dominated within the current view.

DATASETSYN-PPA-2026Q38 fictional runs · frozen demo
COST AXIS
VALUES
PORTABLE EXPERIMENT · V1.vlsilab.jsonLOCAL FILE · NO UPLOAD
Recent

Carries filters, axes, value mode, and selected synthetic run. The frozen benchmark dataset is never duplicated.

8 points visible4 Pareto candidates

28 nm45 nm65 nmP within-workload Pareto

Every point is a synthetic demo run. Select a point to open provenance details. Pareto candidates are marked P.

Current within-workload Pareto candidates
  • Haetae-RV32 · 28 nm
  • Goryeo-AES · 28 nm
  • Nuri-Tensor · 28 nm
  • Dokkaebi-NoC · 45 nm

Pareto status is computed only against visible points in the same workload. It does not demonstrate technical superiority.

VISIBLE SET / 02

Synthetic within-workload comparison

PPA index = performance ÷ √(area × power), with each family baseline = 100. Ordinals reset for every workload and never form a global product ranking.

Synthetic PPA demo comparisons grouped and ranked within workload
GROUP #Design / runWorkloadNodeTarget clockAreaPowerPPA idxSource
CPU core · ranked only inside this workload
01CPU core28 nm780 MHz0.91 mm²34.8 mW138SYNTHETIC
02CPU core45 nm590 MHz1.36 mm²41.2 mW100SYNTHETIC
Crypto · ranked only inside this workload
01Crypto28 nm1,180 MHz0.43 mm²19.7 mW150SYNTHETIC
02Crypto45 nm810 MHz0.74 mm²24.6 mW100SYNTHETIC
AI MAC array · ranked only inside this workload
01AI MAC array28 nm690 MHz3.14 mm²184 mW155SYNTHETIC
02AI MAC array65 nm340 MHz7.38 mm²258 mW100SYNTHETIC
Interconnect · ranked only inside this workload
01Interconnect45 nm930 MHz2.08 mm²67.5 mW140SYNTHETIC
02Interconnect65 nm510 MHz3.96 mm²92.4 mW100SYNTHETIC

Clock, area, power, runtime, and PPA index in every row are generated for UI demonstration. Do not cite them or use them as real process performance. This Atlas is not an EDA sign-off tool or investment advice.

PUBLICATION GATE / 03

Publish evidence before numbers.

01Do not publish observed values without verified PDK and library conditions.

02Do not rank results without a reproducible flow commit and constraints.

03Always separate synthetic examples visually from observed data.