Metraipolis › archive › dc_28dcef33027f
Benchmarking Confidential Computing Performance on NVIDIA Blackwell GPUs — arXiv:2608.26575
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.26575
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T15:28:23.770Z
- SHA-256 of the archived text
3de83f9b6609d0e17f9babab79af9077a08d006abad769bbbd29c0fc1bbcabc4- Stable URL of this page
https://metraipolis.com/documento/dc_28dcef33027f
How to cite
Benchmarking Confidential Computing Performance on NVIDIA Blackwell GPUs — arXiv:2608.26575 — arXiv (en), https://arxiv.org/abs/2608.26575 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_28dcef33027f (Metraipolis archive dc_28dcef33027f, sha256 3de83f9b6609d0e17f9babab79af9077a08d006abad769bbbd29c0fc1bbcabc4)
BibTeX
@misc{dc28dcef33027f,
title = {Benchmarking Confidential Computing Performance on NVIDIA Blackwell GPUs — arXiv:2608.26575},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.26575},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_28dcef33027f; sha256 3de83f9b6609d0e17f9babab79af9077a08d006abad769bbbd29c0fc1bbcabc4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26575. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
This paper measures the performance impact of running large language model inference and training inside a Trusted Execution Environment (TEE) on NVIDIA B200 GPUs, using Intel Trust Domain Extensions (TDX) confidential VMs together with NVIDIA Confidential Computing (CC) on Blackwell GPUs. The performance impact is derived from paired confidential versus non-confidential runs on a single physical host where the only variable is the GPU CC bit and the TDX guest object in the VM launch. The main result is that confidential inference on Blackwell achieves low single-digit throughput overhead when the stack is configured correctly, at about 1-3%. Stock inference stacks incur 30 to 40% penalties due to avoidable configurations rather than the achievable operating point. The cost is not fully represented by a single number because it is governed by two independent axes, a fixed per-host-operation cost that amortizes as batch size grows and a per-NVLink-traffic cost that tracks the share of the step spent in encrypted collectives, and which of the two dominates is set by the workload and the software. We localize each cost to a specific encrypted boundary, give a microbenchmark that predicts the serving penalty to within a submission count, and end with concrete deployment guidance. GPU compute, energy draw, and usable memory capacity are unaffected by CC. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-08-28 in cs.DC). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Cryptography and Security (cs.CR). Pagina del lavoro: https://arxiv.org/abs/2608.26575 — 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 2026-10-17T15:17:44.895Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).