You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[doc-only] Document the CUDA 12 NVML bindings API - #3527
Add the NVML module to the CUDA 12 bindings API navigation and document its functions, types, and exceptions. The reference covers the renamed exceptions and their compatibility aliases, along with the NVML changes called out in the 12.9.11 notes.
I compared the CUDA 12 and CUDA 13 API toctrees and the existing CUDA 13 NVML catalog. CUDA 12 uses a catalog checked against its own installed exports, so each major can document the APIs it ships.
Validation
A full CUDA 12 Sphinx build against the published 12.9.11 wheel created 680 NVML detail pages and search objects. The versioned inventory contains 949 NVML entries, including the renamed exceptions and aliases. The same rendered checks caught the missing navigation and inventory in the baseline build. Both builds reported the same warning count, with zero NVML warnings.
I built the baseline and this PR's CUDA 12 sources with Sphinx 8.1.3 on CPython 3.12.16 and the published cuda-bindings==12.9.11 wheel. The CUDA 12 NVML implementation matches the release tag. The same rendered check failed on the baseline's missing NVML navigation and passed on the candidate.
The API reference links to NVML and renders all 680 catalog detail pages and search objects. The versioned Sphinx inventory contains 949 NVML entries.
The renamed memory and timeout exceptions and their compatibility aliases resolve to their generated pages. The NvlinkFirmwareInfo class page lists firmware_version.
Both builds reported 81 existing warnings and zero NVML warnings. Local all-files and targeted pre-commit checks passed.
No actionable comments were generated in the recent review.
Summary
Summary by CodeRabbit
Documentation
Added an NVML API reference page covering its functions, enums, types, and exceptions.
Added the NVML reference to the API documentation navigation.
Documented compatibility aliases for timeout and memory errors.
The CUDA 12 API reference now includes an NVML documentation page. The page lists NVML functions, enums, types, and exceptions, and identifies TimeoutError and MemoryError as compatibility aliases.
Changes: CUDA 12 NVML documentation
Layer / File(s)
Summary
Add the NVML API reference cuda_bindings_12/docs/source/api.rst, cuda_bindings_12/docs/source/module/nvml.rst
The API toctree links to the new NVML page. The page lists NVML functions, enums, types, and exceptions, and documents two compatibility aliases.
Assessment against linked issues
Objective
Addressed
Explanation
Add a navigable CUDA 12 NVML API reference with documentation for the NVML surface supported by the package [#3517]
✅
Possibly related issues
#3383: The new CUDA 12 NVML autosummary entries overlap with the issue’s focus on NVML autosummary imports, though the issue mainly concerns other documentation warnings.
Suggested reviewers:mdboom, lijinf2
Change: Other · Severity of issue fixed: Medium
Merge Risk: Minimal · up to 39280
This change adds an NVML API reference page to the CUDA 12 docs and links it from the API navigation. It does not alter runtime behavior, and the reported docs build and CI passed. It is ready to merge.
Architecture risk: Low · up to 39280
The change affects 1 system.
Changed systems:cuda_bindings_12
Architecture concerns
No architecture-level concerns identified.
Review details
Systems and components
observed — cuda_bindings_12 (service) was modified; 2 changed files map to changed impact.
Before / after behavior
observed — Modified behavior in cuda_bindings_12/docs/source/api.rst: The API reference toctree adds module/nvml.
observed — Modified behavior in cuda_bindings_12/docs/source/module/nvml.rst: Added the NVML module reference page, with autosummary inventories for functions, enums, types, and exceptions, plus a declaration that TimeoutError and MemoryError are compatibility aliases for the corresponding NVML exceptions.
- [ ] Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Reviewing files that changed from the base of the PR and between edf1b31 and 3928017.
Files selected for processing: 2
cuda_bindings_12/docs/source/api.rst
cuda_bindings_12/docs/source/module/nvml.rst
Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 10 remain after this review.
This branch has not been deployed
No deployments
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cuda.bindingsEverything related to the cuda.bindings moduledocumentationImprovements or additions to documentation
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
— codex GPT-6 on behalf of Ralf —
Description
Closes #3517.
Add the NVML module to the CUDA 12 bindings API navigation and document its functions, types, and exceptions. The reference covers the renamed exceptions and their compatibility aliases, along with the NVML changes called out in the 12.9.11 notes.
I compared the CUDA 12 and CUDA 13 API toctrees and the existing CUDA 13 NVML catalog. CUDA 12 uses a catalog checked against its own installed exports, so each major can document the APIs it ships.
Validation
A full CUDA 12 Sphinx build against the published 12.9.11 wheel created 680 NVML detail pages and search objects. The versioned inventory contains 949 NVML entries, including the renamed exceptions and aliases. The same rendered checks caught the missing navigation and inventory in the baseline build. Both builds reported the same warning count, with zero NVML warnings.
Checklist