bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 31쪽 중 3쪽
PUs...")
print(" This may take 20–30 seconds...")
evidence_list = collect_gpu_evidence(
nonce=nonce,
no_gpu_mode=False,
ppcie_mode=False
)
print(f" Found {len(evidence_list)} GPU(s) with CC support")
print()
# Step 3: Perform attestation
print("[3/3] Validating attestation evidence...")
print(" • Verifying certificate chains")
print(" • Checking OCSP revocation status")
print(" • Validating firmware measurements (RIM)")
print(" • Verifying cryptographic signatures")
print()
arguments = {
'user_mode': False,
'test_no_gpu': False,
'allow_hold_cert': False,
'driver_rim': None,
'vbios_rim': None,
'service_key': None,
'rim_root_cert': None,
'ocsp_url': None,
'rim_service_url': None,
'ocsp_nonce_disabled': False,
'claims_version': '3.0',
'verbose': False
}
result, response = attest(arguments, nonce, evidence_list)
print()
print("=" * 70)
if result:
print("✅ VERIFICATION SUCCESSFUL")
print("=" * 70)
print()
print(f"This system has {len(evidence_list)} genuine NVIDIA GPU(s)")
print("with Confidential Computing features enabled and operational.")
print()
print("Verification completed:")
print(" ✅ Certificate chains are valid")
print(" ✅ Certificates are not revoked (OCSP)")
print(" ✅ Firmware measurements match golden RIM values")
print(" ✅ Hardware attestation passed")
print("=" * 70)
return True
else:
print("❌ VERIFICATION FAILED")
print("=" * 70)
print()
print("This system may not have genuine Confidential Computing GPUs")
print("or CC features may not be correctly configured.")
print()
print("Contact support for assistance.")
print("=" * 70)
return False
except Exception as e:
error_msg = str(e)
# Known non-critical timeout after successful verification
if (
"nvmlSystemSetConfComputeGpusReadyState" in error_msg or
"terminal state" in error_msg
):
print()
print("=" * 70)
print("✅ VERIFICATION SUCCESSFUL")
print("=" * 70)
print()
print(f"This system has {len(evidence_list)} genuine NVIDIA GPU(s)")
print("with Confidential Computing features enabled and operational.")
print()
print("Verification completed:")
print(" ✅ Certificate chains are valid")
print(" ✅ Certificates are not revoked (OCSP)")
print(" ✅ Firmware measurements match golden RIM values")
print(" ✅ Hardware attestation passed")
print()
print("Note: GPU ready-state configuration timed out (non-critical).")
print("=" * 70)
return True
print()
print("=" * 70)
print("❌ VERIFICATION ERROR")
print("=" * 70)
print(f"Error: {error_msg}")
print()
print("Contact support for assistance.")
print("=" * 70)
return False
if __name__ == '__main__':
success = main()
sys.exit(0 if success else 1)
EOF
chmod +x ~/gpu-verification/verify_gpu.py
```
</Expandable>
</Step>
<Step title="Verify the Installation">
Confirm that the NVIDIA attestation packages are installed correctly.
```bash theme={null}
# Check that NVIDIA packages are installed
pip list | grep nv-
```
**Expected output:**
```
nv-attestation-sdk 2.6.3 (or higher)
nv-local-gpu-verifier 2.6.3 (or higher)
```
</Step>
<Step title="Setup Complete" icon="check">
You have successfully installed the official NVIDIA GPU attestation tools.
**What you have created:**
* **Directory:** `~/gpu-verification/`
* **Virtual environment:** `~/gpu-verification/venv/`
* **Verification script:** `~/gpu-verification/verify_gpu.py`
* **Installed packages:** `nv-attestation-sdk`, `nv-local-gpu-verifier`
This setup is **required only once**. The verification script is now ready for repeated use whenever GPU attestation is required.
</Step>
</Steps>
## Running GPU Verification
After completing the one-time setup, you can verify your GPUs at any time by following the steps below.
### Quick Verification
This verification process will take approximately 30 seconds.
```bash theme={null}
# Connect to your virtual machine
ssh root@your-vm-ip
# Navigate to the verification directory
cd ~/gpu-verification
# Activate the Python virtual environment
source venv/bin/activate
# Run the verification script
python3 verify_gpu.py
```
### What Happens During Verification
The verification script performs the following steps automatically:
<Steps>
<Step title="Generating a Cryptographic Nonce">
* Creates a random 32-byte challenge value.
* Ensures attestation freshness and prevents replay attacks.
</Step>
<Step title="Collecting Evidence from GPUs">
* Queries each GPU for attestation evidence (typically takes approximately 20 seconds).
* Retrieves certificate chains directly from the GPU hardware.
* Collects firmware measurements for both the driver and VBIOS.
* Gathers all required cryptographic signatures.
</Step>
<Step title="Validating Attestation Evidence">
* Verifies that certificate chains trace back to the NVIDIA Root Certificate Authority.
* Checks certificate revocation status using OCSP (Online Certificate Status Protocol).
* Retrieves Reference Integrity Manifests (RIMs) from NVIDIA.
* Compares runtime firmware measurements against known golden RIM values.
* Validates all cryptographic signatures to ensure integrity and authenticity.
</Step>
<Step title="Expected Output (Success)" icon="check">
```
======================================================================
NVIDIA GPU Confidential Compute Verification
Using official NVIDIA nv-attestation-sdk
======================================================================
[1/3] Generating cryptographic nonce...
Nonce: 6040f42cc179...
[2/3] Collecting evidence from GPUs...
This may take 20-30 seconds...
Found 8 GPU(s) with CC support
[3/3] Validating attestation evidence...
• Verifying certificate chains
• Checking OCSP revocation status
• Validating firmware measurements (RIM)
• Verifying cryptographic signatures
======================================================================
✅ VERIFICATION SUCCESSFUL
======================================================================
This system has 8 genuine NVIDIA GPU(s)
with Confidential Computing features enabled and operational.
Verification completed:
✅ Certificate chains are valid
✅ Certificates are not revoked (OCSP)
✅ Firmware measurements match golden RIM values
✅ Hardware attestation passed
======================================================================
```
</Step>
</Steps>
### What This Proves
When the output displays **✅ VERIFICATION SUCCESSFUL**, you have cryptographic assurance that:
1. **Authenticity:** All GPUs are genuine NVIDIA hardware with valid certificate chains.
2. **Integrity:** Firmware measurements match NVIDIA’s reference values, indicating no tampering.
3. **Configuration:** Confidential Computing features are enabled and functioning correctly.
4. **Trust Chain:** All certificates trace back to the NVIDIA Root Certificate Authority and have not been revoked.
## Understanding Verification Results
### Successful Verification Indicators
Your GPUs are successfully verified when all of the following messages appear:
* `✅ “VERIFICATION SUCCESSFUL”`
* `✅ “Certificate chains are valid”`
* `✅ “Certificates are not revoked (OCSP)”`
* `✅ “Firmware measurements match golden RIM values”`
* `✅ “Hardware attestation passed”`
**What this confirms:**
* The GPUs are authentic NVIDIA hardware.
* Firmware is intact, unmodified, and matches NVIDIA’s reference measurements.
* Confidential Computing features are enabled and operating correctly.
* The system is suitable for running sensitive workloads.
### Failed Verification Indicators
Contact support immediately if any of the following messages appear:
* `❌ “VERIFICATION FAILED”`
* `❌ “Certificate validation failed”`
* `❌ “RIM verification failed”`
* `❌ “OCSP check failed (revoked)”`
* `❌ “No GPUs with Confidential Computing support found”`
**What this may indicate:**
* The hardware does not support Confidential Computing.
* The system is misconfigured.
* Network connectivity issues are preventing access to OCSP or RIM services.
* A potential security issue that requires further investigation.
<Warning>
**Do not process sensitive or confidential data if verification fails.**
All failures should be investigated and resolved before running protected workloads.
</Warning>
## Troubleshooting
This section outlines common issues encountered during GPU attestation and provides guidance on how to diagnose and resolve them.
<AccordionGroup>
<Accordion title="Issue: "No GPUs with Confidential Computing Support Found"" icon="circle-exclamation">
**Possible causes:**
* The virtual machine type does not support Confidential Computing.
* NVIDIA drivers are not installed or are not properly loaded.
* Confidential Computing features are not enabled at the BIOS or firmware level.
**Verify GPU presence:**
```bash theme={null}
nvidia-smi
```
**Expected result:**\
The output should list NVIDIA H100 or H200 GPUs.
**If no GPUs are shown:**\
Contact support, as the virtual machine may not be provisioned with the correct hardware.
</Accordion>
<Accordion title="Issue: Network Timeouts or Connection Errors" icon="circle-exclamation">
**Possible causes:**
* Firewall rules blocking outbound HTTPS traffic to NVIDIA services.
* No internet connectivity from the virtual machine.
* Temporary unavailability of NVIDIA attestation services.
**Test connectivity to NVIDIA services:**
```bash theme={null}
# Test NVIDIA OCSP server
curl -v https://ocsp.nvidia.com
# Test NVIDIA RIM service
curl -v https://rim.attestation.nvidia.com
```
**If access is blocked:**\
Update firewall rules to allow outbound HTTPS traffic (port 443) to the following endpoints:
* `ocsp.nvidia.com`
* `rim.attestation.nvidia.com`
</Accordion>
<Accordion title="Issue: Verification Takes Longer Than Expected" icon="circle-exclamation">
**Normal execution time:**
* **First run:** 45–60 seconds (initial download of RIM files)
* **Subsequent runs:** 30–45 seconds (cached RIM files are reused)
**If execution exceeds five minutes:**
* Network latency may be affecting access to NVIDIA services.
* Cancel the process using `Ctrl + C` and retry.
* If the issue persists, contact support for further assistance.
</Accordion>
<Accordion title="Issue: “ModuleNotFoundError” or Import Errors" icon="circle-exclamation">
**Possible cause:**\
The Python virtual environment is not activated.
**Solution:**
```bash theme={null}
cd ~/gpu-verification
source venv/bin/activate # (venv) should appear in the prompt
python3 verify_gpu.py
```
</Accordion>
<Accordion title="Issue: Package Installation Fails" icon="circle-exclamation">
**Possible causes:**
* No internet connectivity during setup.
* Inability to reach the PyPI repository.
* Insufficient available disk space.
**Solution:**
```bash theme={null}
# Verify internet connectivity
ping -c 3 pypi.org
# Check available disk space
df -h
# Retry installation
cd ~/gpu-verification
source venv/bin/activate
pip install --upgrade pip
pip install nv-attestation-sdk
```
</Accordion>
</AccordionGroup>
## Support
### Getting Help
If you encounter issues during GPU verification, follow the steps below to diagnose and resolve the problem.
**Recommended Troubleshooting Steps**
1. **Review this guide**\
Most common issues and resolutions are documented in the troubleshooting section.
2. **Examine error messages carefully**\
Error output typically indicates the underlying cause of the failure.
3. **Verify prerequisites**\
Ensure that the NVIDIA driver is installed correctly and that GPUs are visible to the system.
4. **Collect diagnostic information**\
Gather the following information before contacting support:
```bash theme={null}
# Capture GPU information
nvidia-smi > gpu_info.txt
# Capture verification output
python3 verify_gpu.py 2>&1 | tee verification_output.txt
# Capture installed NVIDIA package versions
pip list | grep nv- > package_versions.txt
```
### Contact Support
When reaching out for assistance, provide the Virtual machine identifier, complete error messages, and the diagnostic files listed above.
### NVIDIA Resources
**Official NVIDIA Documentation**
* GPU Attestation: [https://docs.nvidia.com/attestation/](https://docs.nvidia.com/attestation/)
* Confidential Computing: [https://docs.nvidia.com/confidential-computing/](https://docs.nvidia.com/confidential-computing/)
* SDK Repository: [https://github.com/NVIDIA/nvtrust](https://github.com/NVIDIA/nvtrust)
**NVIDIA Support Channels**
* Developer Forums: [https://forums.developer.nvidia.com/](https://forums.developer.nvidia.com/)
* NGC Support: [https://ngc.nvidia.com/support](https://ngc.nvidia.com/support)
# Confidential Compute Attestation
Source: https://io.net/docs/guides/clouds/confidential-compute-attestation-overview
Learn what confidential compute attestation is and how it can be utilized to verify secure workloads on io.net Cloud. Explore Trusted Execution Environments (TEE), GPU verification, and hardware-based security for sensitive compute.
## Overview
[io.net](http://io.net) provisioned *Confidential Compute VMs* are equipped with NVIDIA H200 GPUs featuring hardware-based confidential computing capabilities. This guide shows you how to verify that your GPUs are genuine NVIDIA hardware with confidential computing features properly enabled.
### Key Takeaways
* Understand why GPU verification is essential for confidential computing.
* Learn how to confirm GPU authenticity using cryptographic attestation.
* Follow clear, step-by-step instructions to run verification on your VM.
### Prerequisites
* SSH access to your Confidential Compute VM
* Python 3.7 or later (pre-installed on your VM)
* Basic command line understanding
* 5 minutes to perform the initial set up
## Why does GPU verification matter?
### Trust, but Verify
When running sensitive workloads on **Confidential Computing (CC)** infrastructure, it is not enough to rely on a provider’s assurances. You need **cryptographic proof** that the hardware is authentic, correctly configured, and operating in a secure state.
GPU attestation provides independent verification that:
* **The GPUs are genuine NVIDIA hardware** - not counterfeit, emulated, or misrepresented.
* **Firmware is intact and unmodified** - ensuring no tampering with GPU software or drivers.
* **Confidential computing features are enabled** - confirming that CC mode is active and functioning properly.
* **Hardware measurements match expected “golden” values** - validating that the GPU’s state aligns with NVIDIA’s reference integrity manifests.
### Security and Compliance
GPU attestation is a foundational security mechanism for confidential computing environments. It helps ensure:
* **Zero-trust architecture** - trust no component by default, verify every claim cryptographically.
* **Regulatory and compliance adherence** - meet requirements that mandate hardware verification.
* **Data protection** - guarantee that sensitive workloads run only on validated, trustworthy hardware.
* **Clear auditability** - produce cryptographic evidence for stakeholders, security teams, and auditors.
### How It Works
Attestation relies on **cryptographic proofs** that cannot be falsified, forged or altered. The verification process follows a clear chain of checks:
```
Your VM → Collect GPU Evidence → Verify Certificates
→ Check Measurements
→ Validate Signatures
→ Result: Verified or Check Failed
```
**What gets verified:**
During attestation, the following elements are validated:
* **A four-level GPU certificate chain**, traced back to the NVIDIA Root Certificate Authority.
* **Certificate revocation status**, checked via OCSP to ensure no certificates have been revoked.
* **Driver firmware measurements**, consisting of 64 SHA-384 cryptographic hashes.
* **VBIOS firmware measurements**, consisting of 64 SHA-384 cryptographic hashes.
* **Digital signatures on all attestation evidence**, ensuring integrity and authenticity.
## What is GPU Attestation?
### Cryptographic Proof of Authenticity
GPU attestation is a cryptographic verification process rooted in **hardware-based trust**. It provides verifiable proof that a GPU is authentic, securely configured, and operating in a trusted state. Specifically, it validates four core properties:
<Tabs>
<Tab title="1. Hardware Authenticity">
Every NVIDIA GPU that supports Confidential Computing includes:
* A **unique hardware identity** permanently embedded in silicon during manufacturing.
* A **certificate chain** signed by NVIDIA’s Root Certificate Authority.
* **Cryptographic keys** stored in tamper-resistant hardware.
**What this proves:**\
The GPU is genuine NVIDIA hardware, produced by NVIDIA, and not counterfeit or emulated.
</Tab>
<Tab title="2. Firmware Integrity">
During attestation, the system:
* Measures all GPU firmware using **SHA-384 cryptographic hashes.**
* Compares measurements against NVIDIA’s **Reference Integrity Manifests (RIMs)**.
* Verifies the **digital signatures** on the RIM files.
**What this proves:**\
The GPU firmware has not been modified, tampered with, or compromised.
</Tab>
<Tab title="3. Configuration State">
The verification process confirms that:
* **Confidential Computing (CC)** is enabled.
* **Protected PCIe (PPCIE)** is correctly configured.
* The GPU is operating in the **expected secure state**.
<Note>
In this context, **CC** refers to the general concept of **Confidential Compute**, where the GPU operates in a secure, protected environment. This is different from **CC mode** (also referred to as `CC State`), which is a specific configuration within the GPU.
`CC State` cannot be enabled at the same time as **PPCIe**, the two modes are mutually exclusive.
`CC State` specifically refers to scenarios where a single GPU (or multiple GPUs that are **NOT** NVLink-connected) is passed through directly to a virtual machine.
</Note>
**What this proves:**\
Confidential Computing features are actively enabled and functioning, not merely claimed.
<Accordion title="Further Reading on CC and PPCIe">
**How CC and Protected PCIe (PPCIe) Work Together**
GPU confidentiality is controlled through **two independent mechanisms**:
* **CC (Confidential Compute / CC State)**
* **PPCIe (Protected PCIe)**
These settings determine *how* and *which* GPUs operate in confidential mode. Each GPU can have CC and PPCIe enabled or disabled independently, but only certain combinations are valid.
* **CC = OFF, PPCIe = OFF**\
No confidential computing is enabled.
* **CC = ON, PPCIe = OFF**\
Only a subset of GPUs is confidential. This mode is **NOT compatible with NVLink-connected GPUs**.
* **CC = OFF, PPCIe = ON**\
The **entire GPU set in the server, including NVLink connected GPUs**, operate in confidential mode.
<Note>
*This is the configuration used in our recommended setup.*
</Note>
* **CC = ON, PPCIe = ON**\
This is an invalid configuration. The virtual machine will either fail to boot or the CUDA drivers will not load.
</Accordion>
</Tab>
<Tab title="4. Freshness">
Attestation uses a **nonce** (a cryptographic challenge) to:
* Prevent replay attacks.
* Ensure the evidence is current, not reused or pre-recorded.
* Confirm that measurements reflect the GPU’s **current state**.
**What this proves:**\
The attestation is happening in real time on this specific GPU.
</Tab>
</Tabs>
### Official NVIDIA Technology
**GPU attestation is performed using NVIDIA’s official tooling:**
* **Package:** `nv-attestation-sdk` (official NVIDIA Python SDK)
* **Source:** PyPI (Python Package Index)
* **Version:** 2.6.3 (as of December 2025)
* **License:** Apache 2.0
* **Repository:** [https://github.com/NVIDIA/nvtrust](https://github.com/NVIDIA/nvtrust)
This is **not a third-party tool**. It is NVIDIA’s production-ready attestation framework, used by enterprise customers worldwide.
### Standards-Bas프로젝트로 돌아가기