[None][chore] Use affinity rather than CPU count for default build parallelism

[achartier]

Summary by CodeRabbit

Release Notes

• Improvements
  • Build process now intelligently respects system CPU constraints when determining default parallel jobs, resulting in better resource utilization across different environments.
  • Updated build command help text to reflect improved CPU detection behavior.

Description

On some clusters, the build can have its affinity limited to a subset of CPU cores, as well as a portion of RAM. Launching more processes than available CPU cores is wasteful and raises the risk of running into OOM nvcc crashes.

Test Coverage

N/A

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

[achartier]

@CodeRabbit review

[coderabbitai]

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

📝 Walkthrough

Walkthrough

The change introduces CPU affinity awareness to build parallelism defaults by adding a new helper function that retrieves the number of CPUs available to the current process, honoring Linux CPU affinity constraints. This function replaces direct cpu_count() calls in the job count default handling and argument specification, with corresponding help text updates.

Changes

Cohort / File(s)	Summary
CPU Affinity Support scripts/build_wheel.py	Added get_available_cpu_count() function that respects CPU affinity on Linux before falling back to system CPU count. Updated main job_count default and --job_count argument default to use this new function, with corresponding help text refinements.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Description check	❓ Inconclusive	The PR description includes a clear motivation for the changes and addresses the core issue, but lacks a proper PR title with ticket/type format and omits specific test coverage details.	Add a properly formatted PR title following [type] convention (e.g., [chore]) and provide more specific details about test coverage or explain why N/A is appropriate for this infrastructure change.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: using CPU affinity instead of total CPU count for build parallelism defaults.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

📝 Coding Plan for PR comments

• Generate coding plan

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

scripts/build_wheel.py (1)

2-2: Update copyright year to 2026.

The file has been meaningfully modified. As per coding guidelines, the copyright header should reflect the year of the latest meaningful modification.

Proposed fix

-# SPDX-FileCopyrightText: Copyright (c) 2022-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-FileCopyrightText: Copyright (c) 2022-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@scripts/build_wheel.py` at line 2, Update the SPDX header year from
"2022-2025" to "2022-2026" in the top-of-file copyright comment (the line
starting with "# SPDX-FileCopyrightText: Copyright (c) 2022-2025 NVIDIA
CORPORATION & AFFILIATES. All rights reserved."); replace the year range so the
header reflects the latest meaningful modification year 2026.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@scripts/build_wheel.py`:
- Around line 42-52: The get_available_cpu_count function must handle
cpu_count() possibly returning None; update the except-path in
get_available_cpu_count to coerce the result to an int with a safe default
(e.g., 1) so the function always returns an int. Specifically, when catching
AttributeError and calling cpu_count(), treat a None result as 1 (or use max(1,
cpu_count() or 1)) and return that value. Ensure the change is made inside
get_available_cpu_count so downstream callers (e.g., parallel flags) never
receive None.

---

Nitpick comments:
In `@scripts/build_wheel.py`:
- Line 2: Update the SPDX header year from "2022-2025" to "2022-2026" in the
top-of-file copyright comment (the line starting with "# SPDX-FileCopyrightText:
Copyright (c) 2022-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.");
replace the year range so the header reflects the latest meaningful modification
year 2026.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: b146b208-88c3-4864-967d-8883e376164e

📥 Commits

Reviewing files that changed from the base of the PR and between f11eea7 and b64a634.

📒 Files selected for processing (1)

• scripts/build_wheel.py

[achartier]

/bot run

[tensorrt-cicd]

PR_Github #38762 [ run ] triggered by Bot. Commit: 9f44425 Link to invocation

[tensorrt-cicd]

PR_Github #38762 [ run ] completed with state SUCCESS. Commit: 9f44425
/LLM/main/L0_MergeRequest_PR pipeline #30076 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

[achartier]

/bot run

[tensorrt-cicd]

PR_Github #38783 [ run ] triggered by Bot. Commit: 9f44425 Link to invocation

[tensorrt-cicd]

PR_Github #38783 [ run ] completed with state SUCCESS. Commit: 9f44425
/LLM/main/L0_MergeRequest_PR pipeline #30097 completed with status: 'SUCCESS'

CI Report

Link to invocation