-
My understanding of this is that the GPU attestation (via NVIDIA’s attestation SDK / NRAS) would report measurements of the GPU’s firmware, driver, and VBIOS against known-good reference values, plus confirms the confidential compute mode is active
-
Is there a way for the attestation report to extend to runtime, identifying which specific process, container, or model/workload was actually executing inside the CC-enabled GPU at a given point in time?