Reproducing the literature
10 published QCCD papers. For each one we ran its own code and replayed every schedule in our checker, under the paper's own rules and clock, then scheduled the same circuits on the same machine ourselves, under the same rules. For nine of them that schedule comes from a scheduler we wrote for the paper's own model; for Quantinuum H2 it comes from this website's compiler. Open a paper for its machine, every run, what we found, what this website takes from it, and the animations.
| paper | their code | their numbers, replayed | ours on the same machine | |
|---|---|---|---|---|
| QCCDSim ISCA 2020we build on its 3 rules, transport times, chain-length error, grid baseline | no license | 12/12 runs exact 24/24 paper values 2 findings | 7 of 9 faster, 1.06–3.16× | details → |
| Muzzle the Shuttle DATE 2022 | Apache-2.0 | 10/10 runs exact 6/10 paper values 4 findings | fewer shuttles than Muzzle: 0 of 5 faster than QCCDSim: 3 of 5 | details → |
| Cyclone HPCA 2026we build on its 5 rules, rigid rotation, ring device and timing table | MIT | 13/13 runs exact 8/8 paper values 3 findings | 0 of 6 faster | details → |
| Surface-code QCCD ASPLOS 2026we build on its 7 rules, objective, transport heating bounds, starting question | MIT | 17/17 runs exact 7/9 paper values 3 findings | 4 of 4 faster, 1.10–1.19× | details → |
| TISCC SC-W 2023 | UT-Battelle open source license (BSD-style) | 1/2 runs exact 1 finding | 2 of 2 faster, 1.08× | details → |
| Position Graph arXiv 2025 | not published | 1/1 runs exact 0/1 paper values 1 finding | 1 of 1 faster, 1.76× | details → |
| Moveless arXiv 2025 | no license | 12/12 runs exact 2 findings | 2 of 4 faster, 1.03–1.04× | details → |
| Quantinuum H2 Phys. Rev. X 13, 041052 (2023)we build on its 2 rules, race-track device, gate fidelities | circuits public, all rights reserved | machine rebuilt; the published circuits' gate counts match | 248 of 248 circuits compile; fewer rounds than the paper in 4 of 5 families rounds only: our model does not price transport | details → |
| TrapSIMD arXiv 2504.17886 (2025)we build on its 5 rules, movement classes, gate and measurement times | not published | 4/4 worked examples exact no code published | 32 of 32 faster, 1.1–12.8× model level, rebuilt circuits | details → |
| MQT IonShuttler ASP-DAC 2024we build on its SAT-oracle idea | MIT | 75/75 runs exact 3/6 paper values 3 findings | 15 of 15 faster than the heuristic; each at its lower bound under the heuristic's model | details → |
How a paper is checked, and which rules each paper's model has · Papers we read and did not reproduce · every paper's rule set