Complete NVIDIA DGX keyboard shortcuts and commands reference — 16 shortcuts across 3 categories. Quick reference cheat sheet for Windows & Mac.
A DGX system is a server with its own management stack, and this page gathers the commands an operator runs to check it: NVSM for the system as a whole, nvidia-smi and DCGM for the GPUs and NVLink, InfiniBand tools for the fabric, and the kernel log for XID errors. The notes follow the order of a health check, from "is the box healthy" down to "which GPU".
| Shortcut | Action | Description |
|---|---|---|
| nvsm show health | System health | Check overall DGX system health. |
| nvsm show gpu | GPU status | Check status of all GPUs. |
| nvsm show nvswitch | NVSwitch status | Check NVSwitch connection status. |
| nvsm show memory | Memory status | Check system memory status. |
| nvsm dump | Full dump | Dump complete system information. |
| nvidia-smi topo -m | GPU topology | Show GPU NVLink topology. |
| dcgmi diag -r 3 | GPU diagnostics | Run full GPU diagnostics with DCGM (Level 3). |
| dcgmi health -c -g 0 | DCGM health check | Check health of DCGM group 0. |
| Shortcut | Action | Description |
|---|---|---|
| ibstat | IB port status | Check InfiniBand port status. |
| ibstatus | IB status summary | Summarize InfiniBand interface status. |
| nvidia-smi nvlink -s | NVLink status | Check NVLink link status and bandwidth. |
| lspci | grep -i nvidia | PCI GPU check | List NVIDIA devices on the PCI bus. |
| Shortcut | Action | Description |
|---|---|---|
| nvsm show psu | PSU status | Show power supply unit status and redundancy state. |
| nvsm show fans | Fan status | Show fan module speed and health status. |
| dmesg | grep -i xid | XID error check | Search the kernel log for NVIDIA XID GPU error codes. |
| systemctl status nvsm | NVSM service status | Check whether the NVSM monitoring service is running. |
→ Related guide: GPU cluster management: the full command guide
The most essential NVIDIA DGX shortcuts are: nvsm show health (System health), nvsm show gpu (GPU status), nvsm show nvswitch (NVSwitch status).
These are command-line commands — type them in your terminal or console. Combine them with shell history search (Ctrl + R) and aliases to work even faster.
The NVIDIA DGX shortcut for system health is nvsm show health. Check overall DGX system health.
Yes — use My Stack to combine NVIDIA DGX shortcuts with any other platform on this site into one printable reference, which is useful if your daily workflow spans several tools.
nvsm show health runs the system health check and reports each subsystem as healthy or not, and is the first command after any alert. nvsm show gpu, nvsm show nvswitch and nvsm show memory drill into the accelerators, the NVSwitch fabric and DIMMs; nvsm show psu and nvsm show fans cover power and cooling. nvsm dump collects everything into a bundle for NVIDIA Enterprise Support — run it before opening a case. systemctl status nvsm confirms the NVSM service is running when its commands hang.
nvidia-smi topo -m prints the GPU-to-GPU and GPU-to-NIC topology matrix — NVLink, PCIe host bridge, NUMA — which explains why some GPU pairs are faster than others and informs how jobs are placed. nvidia-smi nvlink -s shows every NVLink's state and speed; a link that is inactive on one GPU is a hardware fault. dcgmi diag -r 3 runs the full DCGM diagnostic during maintenance, and dcgmi health -c -g 0 checks the health watches on GPU group 0 between jobs. lspci | grep -i nvidia confirms the OS sees every GPU and NVSwitch, which catches a device that dropped off the PCIe bus.
ibstat and ibstatus show the InfiniBand ports, which should all be Active at the expected rate on a multi-node cluster. dmesg | grep -i xid searches the kernel log for XID errors — the driver's error codes for GPU faults, ECC events and application crashes — and the XID number maps to a documented cause; repeated XIDs on one GPU are the signal to drain the node.
Open your assistant with this page preloaded as the source — great for follow-up questions like "which of these work in other apps?"