Nvidia DGX A100 Service Manual

Chapter 16. Updating the Mellanox
Network Card Firmware
After replacing or installing the Mellanox ConnectX cards, make sure the rmware on the cards is up
to date.
1. Conrm the OS is updated to the latest version; this will ensure the latest tested and supported
rmware is downloaded.
$ sudo apt update
$ sudo apt full-upgrade
2. Run the rmware update script as follows:
$ sudo ∕opt∕mellanox∕mlnx-fw-updater∕mlnx_fw_updater.pl
If the rmware version in the new card is more recent than the one available in the operating
system, the rmware should be downgraded to a supported version by issuing the following:
$ sudo ∕opt∕mellanox∕mlnx-fw-updater∕mlnx_fw_updater.pl --force-fw-update
3. Reboot the system for the rmware update to take eect.
$ sudo reboot
4. Once the system comes up, make sure the rmware versions are all the same by issuing:
$ cat ∕sys∕class∕infiniband∕mlx5_*∕fw_ver
77

Nvidia DGX A100 Specifications

General

GPU	8 x NVIDIA A100 Tensor Core GPUs
System Memory	1 TB DDR4
Storage	15 TB NVMe SSD
GPU Memory	320 GB total (40 GB per GPU)
CPU	2 x 64-Core AMD EPYC 7742
Networking	8 x 200 Gb/s InfiniBand or Ethernet
Interconnect	NVIDIA NVLink