Base nvbandwidth on cuda image

[guptaNswati]

For OSRB compliance, we need to base the nvbandwidth image on a cuda base image.

Single node bandwidth test with the updated base image:

docker run -it -e NVIDIA_IMEX_CHANNELS=0 -e NVIDIA_VISIBLE_DEVICES=all nvbandwidth:0.1 --allow-run-as-root  -n 2 nvbandwidth -p multinode_device_to_device_memcpy_read_ce
nvbandwidth Version: v0.7
Built from Git version: v0.7

MPI version: Open MPI v4.1.2, package: Debian OpenMPI, ident: 4.1.2, repo rev: v4.1.2, Nov 24, 2021
CUDA Runtime Version: 12080
CUDA Driver Version: 12080
Driver Version: 570.86.15

Process 0 (c893d00c6756): device 0: NVIDIA GH200 96GB HBM3 (00000009:01:00)
Process 1 (c893d00c6756): device 1: NVIDIA GH200 96GB HBM3 (00000019:01:00)

Running multinode_device_to_device_memcpy_read_ce.
memcpy CE GPU(row) -> GPU(column) bandwidth (GB/s)
           0         1
 0       N/A    391.93
 1    391.88       N/A

SUM multinode_device_to_device_memcpy_read_ce 783.81

NOTE: The reported results may not reflect the full capabilities of the platform.
Performance can vary with software drivers, hardware clocks, and system topology.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[guptaNswati]

/ok-to-test

Copilot wasn't able to review any files in this pull request.

Files not reviewed (1)

• deployments/container/nvbandwidth/Dockerfile: Language not supported