feat: Added power support for python backend on ubuntu.

[Sunidhi-Gaonkar1]

What does the PR do?

Checklist

• I have read the Contribution guidelines and signed the Contributor License
  Agreement
• PR title reflects the change and is of format <commit_type>: <Title>
• Changes are described in the pull request.
• Related issues are referenced.
• Populated github labels field
• Added test plan and verified test passes.
• Verified that the PR passes existing CI.
• I ran pre-commit locally (pre-commit install, pre-commit run --all)
• Verified copyright is correct on all changed files.
• Added succinct git squash message before merging ref.
• All template sections are filled out.
• Optional: Additional screenshots for behavior/output changes with before/after.

Commit Type:

Check the conventional commit type
box here and add the label to the github PR.

• build
• ci
• docs
• feat
• fix
• perf
• refactor
• revert
• style
• test

Related PRs:

Where should the reviewer start?

Test plan:

Used below commands to create triton server docker instance and starting up the triton server:
1.sudo docker run --privileged -it -p8000:8000 --volume /model_repository:/var/models --name triton_server_test tritonserver
2../bin/tritonserver --model-repository=/var/models

Tested the simple identity model using below commands in a different session:
1.curl -X POST localhost:8000/v2/repository/index | jq

curl -X POST localhost:8000/v2/repository/index | jq
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100    58  100    58    0     0  29000      0 --:--:-- --:--:-- --:--:-- 29000
[
  {
    "name": "simple_identity",
    "version": "1",
    "state": "READY"
  }
]

2.curl localhost:8000/v2/models/simple_identity | jq

curl localhost:8000/v2/models/simple_identity | jq
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   193  100   193    0     0   188k      0 --:--:-- --:--:-- --:--:--  188k
{
  "name": "simple_identity",
  "versions": [
    "1"
  ],
  "platform": "python",
  "inputs": [
    {
      "name": "INPUT0",
      "datatype": "BYTES",
      "shape": [
        -1,
        -1
      ]
    }
  ],
  "outputs": [
    {
      "name": "OUTPUT0",
      "datatype": "BYTES",
      "shape": [
        -1,
        -1
      ]
    }
  ]
}

Caveats:

Background

Hi team,
This PR includes changes made to add Power support to triton-inference-server for python backend. Can you please review/merge this?

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

• closes GitHub issue: #xxx

[mkumatag]

I feel that the cpu variant could be applicable for other use cases as well, instead of limiting it to the ppc64le architecture. Can we just parameterize this variant and set all as a default value (this way we can safely remove this uname check here)?

[Sunidhi-Gaonkar1]

Sure, we are working on these changes.

[Sunidhi-Gaonkar1]

I have made the changes to remove uname check , added a --build_variant flag which is set to all by default.

[mkumatag]

why are we installing explicitly on ppc64le architecture and not on other architectures?

[Sunidhi-Gaonkar1]

We were facing linker issue for numpy as it was getting built from source. Hence, need to install clang-15 for ppc64le architecture. For other architectures numpy wheels are available.

[Sunidhi-Gaonkar1]

I have removed these changes as we are no longer facing the linker issue.