Metraipolis › archive › dc_fc53383e6e1e

Matching Workloads to Blackwell: Power-Resolved Benchmarking of NVIDIA B200 and B300 across HPC and AI Workloads — arXiv:2610.41919

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41919
License (as archived)
CC0-1.0
Archived
2026-11-03T08:07:21.210Z
SHA-256 of the archived text
5a9a63ab9eb5acb2e068586dda33ebe484e6c2091e469d6ee49f6850f0f10b07
Stable URL of this page
https://metraipolis.com/documento/dc_fc53383e6e1e

How to cite

Matching Workloads to Blackwell: Power-Resolved Benchmarking of NVIDIA B200 and B300 across HPC and AI Workloads — arXiv:2610.41919 — arXiv (en), https://arxiv.org/abs/2610.41919 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fc53383e6e1e (Metraipolis archive dc_fc53383e6e1e, sha256 5a9a63ab9eb5acb2e068586dda33ebe484e6c2091e469d6ee49f6850f0f10b07)
BibTeX
@misc{dcfc53383e6e1e,
  title = {Matching Workloads to Blackwell: Power-Resolved Benchmarking of NVIDIA B200 and B300 across HPC and AI Workloads — arXiv:2610.41919},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41919},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fc53383e6e1e; sha256 5a9a63ab9eb5acb2e068586dda33ebe484e6c2091e469d6ee49f6850f0f10b07},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41919. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We evaluate NVIDIA B200 and B300 (Blackwell Ultra) GPUs on two independent eight-GPU clusters, spanning computer-vision inference and training, LLM inference with Llama-3-8B and 70B~\cite{llama3}, and six production HPC codes. The B300 is not a uniform successor to the B200: it trails by up to 5.6$\times$ on convolutional inference, 8$\times$ on 3D vision transformer training, 24--27\% on 70B LLM throughput, and 16$\times$ on double-precision Quantum ESPRESSO. We resolve the convolutional deficit using per-GPU board power rather than throughput alone: on the affected workloads the B300 draws only 36--53\% of its 1100\,W envelope, whereas on ViT-Large, which it wins, both parts saturate at 95--97\% of TDP. Because our harness keeps inputs resident in device memory, a host-side data pipeline is excluded by construction, localizing the cause to per-kernel cost. Two B300 nodes of identical silicon differ by 2.1$\times$ in time-to-first-token solely from a development vLLM build. We give per-workload procurement guidance.

Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.DC, nuovi lavori del giorno).
Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Hardware Architecture (cs.AR).
Pagina del lavoro: https://arxiv.org/abs/2610.41919 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T08:00:08.597Z dalla pagina https://arxiv.org/list/cs.DC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).