Panel: A New Root Of Trust: Changing Computing With TEE
summary
Panel gives a concrete account of trusted execution: CPU-built TEEs (Intel SGX/TDX, ARM CCA Realms, AMD SEV SNP, Nvidia H100) encrypt runtime memory and produce signed measurement reports for attestation, with nonces to prevent replay. Vendors publish TDX module and SEAM loader source; full open source remains limited. Early implementations hit 50–100% slowdowns but modern memory encryption is under 1%; device-to-TE TDX connect promises faster GPU IO. Protections exclude invasive physical decapping and some DoS vectors.
AI-written summary · about
chapters (11)
- 0:00Panel introduction
- 1:29TEE vs TPM explained
- 5:22What is attestation
- 14:09Supply chain trust model
- 17:20Confidential AI use case
- 20:59TEE performance impact
- 23:47Firmware vs software supply chains
- 27:01Open source in TEEs
- 31:22Replay attacks in attestation
- 33:27TEE security boundaries
- 38:43Q&A session
related talks
37mIt's All About Trust: Why Confidential Computing Matters
42mSponsored Session: Confidential Computing - New Capabilities for New Workloads
36mEverybody is talking about Confidential Computing - the minimum Devs should know- 42mSponsored Session: Confidential Computing - New Capabilities for New Workloads
39mSoftware Supply Chain Security with Trusted Execution Environment Provisioning (T...
47mThe Rise of Confidential Computing, What It Is and What it Means to You- 37mIt's All About Trust: Why Confidential Computing Matters
31mIntel Trust Domain Extensions