Records · kv:2610.00027v1

Reproduction of Broadcom/csg-htsim: NDP sequential all-to-all on a 1024-node fat tree

K-Veritas Team

Submitted by K-Veritas Team

Published 2026-10-08

Independent reproduction · Artifact evaluation · Networking

Reproduction of: original code

Abstract

Built htsim at the pinned commit and ran the shipped experiment experiments/fattree-all-to-all/sequential_512conns: a 512-host sequential (shift) all-to-all with NDP and per-packet spraying on a 1024-node three-tier fat tree, 1 MB flows, seed 13, using the repository's own run.sh and post-processing scripts. The result of interest is the self-synchronization described in the experiment's README: how many senders converge on a single receiver. The final gnuplot step was skipped (gnuplot not installed); the older example programs under sim/datacenter/old_tests do not compile at this commit and were not used. Single-threaded simulation on an Intel Xeon W-2145.

Sealed report

Cite this record

@misc{kveritas261000027v1,
  title={Reproduction of Broadcom/csg-htsim: NDP sequential all-to-all on a 1024-node fat tree},
  author={K-Veritas Team},
  year={2026},
  howpublished={K-Veritas Records},
  note={kv:2610.00027v1},
  url={https://kveritas.org/records/2610.00027v1},
}

References

1. [1] Mamadou K. Keita and Christopher Homan. Computer Science Conferences Should Require Nonrepudiable Experimental Results. NeurIPS 2026 Position Paper Track. arXiv:2605.08586.

Sealed with K-Veritas [1]. Listed, not endorsed. A record shows these results came from the sealed code, unchanged. Whether the work is sound is for the reader to judge.