Hi forums !
I’m using a DGX-1(V100) for my deeplearning framworks.
I am doing a test in NVCaffe (docker image, v 17.10) but I found a issue about NVLink throughput
I could think 'nvidia-smi link -g 0 -i ’ command can measure nvlink stream but It does not work !
I already checked nvlink status of all of gpus in DGX-1 system and gpus’ status is all activated.
when I type ‘nvidia-smi nvlink -g 0 -i 0’, then I can only see
Link 0: Rx0: 0 Kbytes, Tx0: 0 Kbytes
Link 1: Rx0: 0 Kbytes, Tx0: 0 Kbytes
Link 2: Rx0: 0 Kbytes, Tx0: 0 Kbytes
Link 3: Rx0: 0 Kbytes, Tx0: 0 Kbytes
Link 4: Rx0: 0 Kbytes, Tx0: 0 Kbytes
Link 5: Rx0: 0 Kbytes, Tx0: 0 Kbytes
that’s all.
Judging by seeing the performance results, it seems to be using NVLink. ( Acutally I did not edit NVCaffe code ! )
Maybe I do not know how to use the command.
Please Help me.