Nik Konyuchenko
Nik Konyuchenko
Cuda-12 will be supported in >=3.1.5 version. We are going to release it at the beginning of this week. On Mon, Dec 12, 2022 at 12:26 AM ligeweiwu ***@***.***> wrote:...
@lahwaacz, Could you try the solution mentioned here? https://github.com/containers/buildah/issues/1702#issuecomment-508143700
@AltarIbnL, The DCGM 1.3.7 provided the DcgmProfTesterKernels.ptx compiled with Cuda12 SDK accidentally. As a short-term solution, you could build the ptx file yourself using the buildptx.sh script and change the...
@AltarIbnL, You should also change `__double2half` to `__float2half` in the DcgmProfTesterKernels.cu file (two instances next to each other), as Cuda10 didn't know about `__double2half`.
@AltarIbnL, We are sorry to inform you that we only offer support for Nvidia GPUs. Unfortunately, since your GPU is not an Nvidia product, we apologize for being unable to...
@bergentruckung, That is kind of expected on MIG-enabled GPUs. The metrics `dcgmi stats` uses do not work for the MIG devices. Namely, those are DCGM_FI_DEV_GPU_UTIL, DCGM_FI_DEV_MEM_COPY_UTIL, DCGM_FI_DEV_PCIE_TX_THROUGHPUT, DCGM_FI_DEV_PCIE_RX_THROUGHPUT. There is...
> @nikkon-dev Is it possible to get per-PID utilization with MIG-enabled GPUs using dcgm? Regrettably, no tool or library offers utilization metrics specific to a PID in MIG mode. The...
@irvingans, The dcgmi/Diag.cpp is just a launcher for the nvvs binary. Please take a look at [this code](https://github.com/NVIDIA/DCGM/blob/7e1012302679e4bb7496483b32dcffb56e528c92/nvvs/plugin_src/software/Software.cpp#L69) The software nvvs plugin checks if any known preconditions may affect/slow down...
Some description is also available [here](https://docs.nvidia.com/datacenter/dcgm/latest/user-guide/dcgm-diagnostics.html#deployment-plugin)