DGX H200 - ECC Memory Features and Power Supply Certification

Accelerated Computing DGX User Forum
1

Hello,

We are evaluating the NVIDIA DGX H200 system for a customer project, and there are two specific hardware requirements we need to verify before proceeding:

  1. Memory Reliability Features
    The customer’s technical specification requires that the server memory support advanced data integrity and fault tolerance mechanisms such as ECC, SDDC, Chipkill, memory mirroring, DIMM sparing, or Memory RAID.
  • Could you please confirm which of these features are supported on the DGX H200’s factory-installed memory modules?
  1. Power Supply Redundancy and Certification
    The system must have redundant, hot-swappable power supplies with 80 Plus Platinum certification.
  • Could you confirm whether the DGX H200’s power supplies meet this certification level?

We would appreciate any official documentation or references you could share that confirm compliance with these requirements.

Thank you in advance for your support.

Best regards,
Ugur