Examining GPU Resources on an ESXi Host

To better manage the GPU resources that are available on an ESXi host, you can examine the current GPU resource reservation. The ESXi command-line query utility, gpuvm, lists the GPUs that are installed on an ESXi host and displays the amount of GPU memory that is reserved for each virtual machine on the host. Note that this GPU memory reservation is not the same as virtual machine VRAM size.

To run the utility, type gpuvm from a shell prompt on the ESXi host. You can use a console on the host or an SSH connection.

For example, the utility might display the following output:

~ # gpuvm
Xserver unix:0, GPU maximum memory 2076672KB
        pid 118561, VM "JB-w7-64-FC3", reserved 131072KB of GPU memory.
        pid 64408, VM "JB-w7-64-FC5", reserved 261120KB of GPU memory.
        GPU memory left 1684480KB.

Similarly, you can use the nvidia-smi command on the ESXi host to see a list of NVIDIA GRID vGPU-enabled virtual machines, the amount of frame buffer memory consumed, and the slot ID of the physical GPU that the virtual machine is using.