This paper presents a direct 1: 1 performance comparison between CIQS, the first fully analytic quantum circuit compilation pipeline, and the five SDKs evaluated in IBM’s Benchpress study: Qiskit, QTS, Tket, BQSKit, and Staq. The paper's approach to the results is two-fold: Under IBM's own published framework, CIQS sits ahead of every SDK in the benchmark: up to fifteen times faster than Qiskit, while scaling beyond one million qubits, three orders of magnitude past where the IBM benchmark ends. Standard benchmark metrics such as 2Q gate count and depth are misleading measures and are insufficient to characterize the output quality of a compiled circuit during real-life operation. Both points are further discussed in the paper's conclusion and supported by external experimental results referenced therein. A complete reference dataset covering all benchmark metrics is provided as ancillary material (benchpresscomparison. pdf). Raw CIQS benchmark results are also included (CIQSbenchpressᵣawᵣesults. json). --------------------------------------- CIQS is freely available on GitHub for non-commercial use: CIQS GitHub page CIQS introduction paper: https: //doi. org/10. 5281/zenodo. 19056796 CIQS ran a black hole: https: //doi. org/10. 5281/zenodo. 19473695 --------------------------------------- arcadialab. fr
Jessy Pensédent (Sat,) studied this question.