Hi, I’m new to the confidential computation field and I’m try verifying the CC capability of Hopper arch on a environment with TDX+H200. I have configured the confidential VM by following the tutorial of the guide (https://docs.nvidia.com/cc-deployment-guide-tdx.pdf). Now, I call the Nvidia attestation sdk API to attest GPU, but get failed due to the send_ocsp_request call timed out. I use the local gpu verifier and find that it still rely on the service of RIM and OCSP, which need network access. However, our environment is air-gapped, I wonder is there any solution for attestation in an offline environment with network access limitation.
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Can I test Confidential Computing with H100 and TDX on Different Machines | 2 | 99 | January 6, 2025 | |
Inquiry on Confidential Computing GPUs | 2 | 567 | February 3, 2024 | |
Confidential Computing Deployment Guide (Intel TDX & KVM) attestation execution fails | 0 | 84 | February 4, 2025 | |
H200 Confidential Compute Fully Supported? | 0 | 121 | March 3, 2025 | |
H100 GPUs in Confidential Computing (CC) mode | 1 | 168 | December 30, 2024 | |
Amazon EC2 GPU Compute servers do not support Compute (in Windows)! | 1 | 1478 | October 17, 2014 | |
GPU authenticity | 1 | 134 | November 25, 2024 | |
Can I run cuda on Virtual machine | 1 | 8334 | April 13, 2018 | |
Discrepancy in the https://docs.nvidia.com/cc-deployment-guide-tdx.pdf | 4 | 89 | September 20, 2024 | |
How can I passthrogh GTX 1080 to virtual machine? | 1 | 2703 | June 6, 2017 |