# Confidential Computing on NVIDIA Hopper GPUs: A Performance Benchmark Study

arXiv: https://arxiv.org/abs/2409.03992 · HTML: https://mlsanity.com/paper/2409.03992/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Builds on 2 of its 16 references that are in the map.
- First posted on arXiv 6 Sep 2024.

## Builds on (most cited first, 2 of 2)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Etalon: Holistic Performance Evaluation Framework for LLM Inference Systems](https://mlsanity.com/paper/2407.07000/) (2024)

## Built on by (most cited first, 8 of 8)

- [CoreGuard: Safeguarding Foundational Capabilities of LLMs Against Model Stealing in Edge Deployment](https://mlsanity.com/paper/2410.13903/) (2024)
- [MemTrust: A Zero-Trust Architecture for Unified AI Memory System](https://mlsanity.com/paper/2601.07004/) (2026)
- [Crypto x AI, AI x Crypto: A Survey](https://mlsanity.com/paper/2606.13892/) (2026)
- [Privacy Failure in Split-LLM Training, The Returned Gradient Nullifies the Decoys](https://mlsanity.com/paper/2609.04382/) (2026)
- [Committed SAE-Feature Traces for Audited-Session Substitution Detection in Hosted LLMs](https://mlsanity.com/paper/2604.18179/) (2026)
- [The Serialized Bridge: Understanding and Recovering LLM Serving Performance under Blackwell GPU Confidential Computing](https://mlsanity.com/paper/2606.23969/) (2026)
- [Benchmarking Confidential GPU Inference on NVIDIA H100 under Intel TDX](https://mlsanity.com/paper/2607.19353/) (2026)
- [SoK: Decentralized Agent Economic Infrastructure](https://mlsanity.com/paper/2610.01756/) (2026)

## Abstract (arXiv)

This report evaluates the performance impact of enabling Trusted Execution Environments (TEE) on NVIDIA Hopper GPUs for large language model (LLM) inference tasks. We benchmark the overhead introduced by TEE mode across various LLMs and token lengths, with a particular focus on the bottleneck caused by CPU-GPU data transfers via PCIe. Our results indicate that while there is minimal computational overhead within the GPU, the overall performance penalty is primarily attributable to data transfer. For the majority of typical LLM queries, the overhead remains below 7%, with larger models and longer sequences experiencing nearly zero overhead.