[SOLVED] When will NVIDIA fix `nvidia-smi -a` ?

You can surely export these things to regular (non Tesla) users, 'cause there are no trade secrets to be exposed by showing this information.

GPU Link Info
            PCIe Generation
                Max                 : N/A
                Current             : N/A
            Link Width
                Max                 : N/A
                Current             : N/A

    Performance State               : N/A

    Clocks Throttle Reasons         : N/A

   Utilization
        Gpu                         : N/A
        Memory                      : N/A

    Power Readings
        Power Management            : N/A
        Power Draw                  : N/A
        Power Limit                 : N/A
        Default Power Limit         : N/A
        Min Power Limit             : N/A
        Max Power Limit             : N/A
    Clocks
        Graphics                    : N/A
        SM                          : N/A
        Memory                      : N/A
    Applications Clocks
        Graphics                    : N/A
        Memory                      : N/A
    Default Applications Clocks
        Graphics                    : N/A
        Memory                      : N/A
    Max Clocks
        Graphics                    : N/A
        SM                          : N/A
        Memory                      : N/A
    Compute Processes               : N/A

They deliberately removed support for non quadro gpus, older drivers (can’t remember the version) worked.

Probably for marketing reason, that’s the drawback of non FOSS drivers, we’ve to deal with it.

Half of this information can be seen using

nvidia-settings -q all

so I guess it’s a case of negligence rather than a deliberate attempt not to expose it.

As i wrote, that info used to work till a specific driver release.
Many users wrote about it, and never nvidia fixed it, nor gave us an internal bug id.

Feel free to wait for an answer :D

I will :-)

Wait until the heat death of the sun… Like so many other bugs.

I’m gonna be nasty.

Bumpy.

Long time, another bump.

Bump.

Bump.

Bump.

A bit impatient there, birdie?

I found the bug - it’s actually inside NVML. I have a bug report and workaround here but as nvidia-smi is closed-source, I can’t apply the bugfix to nvidia-smi without some GDB trickery. :(

If all you want to see is GPU utilization and power usage, well, the code I have there will do just that, provided you have recent drivers. :)

Thank you, sir!

Just improved my bugfix - newer version here gets the mainline nvidia-smi working again.