Document how NVIDIA releases differ from CCCL releases

[wmaxey]

Description

closes

Checklist

• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 9f08cf00-7544-4f93-9375-d90615fe1a9f

📥 Commits

Reviewing files that changed from the base of the PR and between 4b87e7c and c6ae919.

📒 Files selected for processing (1)

• docs/maintainers/branching_strategy.rst

🚧 Files skipped from review as they are similar to previous changes (1)

• docs/maintainers/branching_strategy.rst

---

📝 Walkthrough

Summary by CodeRabbit

• Documentation
  • Documented the ctk/X.Y.x branching model for releases matching exact CUDA Toolkit contents.
  • Clarified NVIDIA ownership, permitted additive internal features, and read-only branch status.
  • Updated release-tagging conventions to use colon separators, including -ctkN.M.K tags.
  • Updated release-candidate tag examples to reflect the standardized separator and format.

Walkthrough

The branching strategy documents NVIDIA-owned, read-only ctk/X.Y.x branches and updates release-tag conventions for CUDA Toolkit release contents.

Changes

Branching strategy

Layer / File(s)	Summary
CUDA Toolkit branches and release tags docs/maintainers/branching_strategy.rst	Documents ctk/X.Y.x branch properties, updates tag separators, and adds the vX.Y.Z-ctkN.M.K release-tag format.

Suggested reviewers: gonidelis

---

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

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

docs/maintainers/branching_strategy.rst (1)

23-24: 🗄️ Data Integrity & Integration | 🔵 Trivial | ⚡ Quick win

suggestion: Clarify that ctk/X.Y.x starts from the exact contents of the corresponding CUDA Toolkit release and may then include additive internally developed features. The current wording can imply that the branch remains an unchanged release snapshot.

Source: Path instructions

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: c29d30b7-4d23-4411-9222-83592481f759

📥 Commits

Reviewing files that changed from the base of the PR and between fd8c64e and 0350d06.

📒 Files selected for processing (1)

• docs/maintainers/branching_strategy.rst

[jrhemstad]

Should also answer when is this branch created? And how?

[github-actions]

⏱️ CCCL compile-time benchmark comparison: Public headers compile-time bench

Result: 0 regression row(s), 26 improvement row(s) above threshold.

Run	Value
Config	public-headers-gcc13
Baseline	origin/main
Preset	all-dev
Targets	cub.headers.base, thrust.cpp.cuda.headers.base, libcudacxx.test.public_headers
GPU / launch args	rtx2080 / --cuda 13.3 --host gcc13

Artifacts: reports and traces

TU total compilation

-f total-compilation inclusive --sort total

🟢 TU total compilation — Improvements

Rank	Improvement impact	Selected Δ	Baseline	Current	Event	Matched traces
1	4.825781	-4.825781	17.830408	13.004627	Total Compilation Time: thrust/unique.h	1
2	3.450788	-3.450788	18.363801	14.913013	Total Compilation Time: cub/cub.cuh	1
3	3.412799	-3.412799	16.689145	13.276346	Total Compilation Time: thrust/device_new_allocator.h	1
4	3.368969	-3.368969	17.666353	14.297384	Total Compilation Time: thrust/partition.h	1
5	3.338721	-3.338721	15.976253	12.637532	Total Compilation Time: thrust/uninitialized_copy.h	1
6	3.320288	-3.320288	10.904631	7.584343	Total Compilation Time: cub/device/dispatch/tuning/tuning_batched_topk.cuh	1
7	3.293683	-3.293683	9.903812	6.610129	Total Compilation Time: thrust/iterator/retag.h	1
8	3.192594	-3.192594	16.789073	13.596479	Total Compilation Time: thrust/equal.h	1
9	3.161839	-3.161839	17.091877	13.930038	Total Compilation Time: thrust/mr/disjoint_sync_pool.h	1
10	3.135457	-3.135457	16.800156	13.664699	Total Compilation Time: thrust/count.h	1
11	3.119875	-3.119875	15.060012	11.940137	Total Compilation Time: thrust/scan.h	1

Direct file processing

-f file-processing exclusive --sort total

🟢 Direct file processing — Improvements

Rank	Improvement impact	Selected Δ	Baseline	Current	Event	Matched traces
1	3.251084	-3.251084	23.293607	20.042523	Processing Header File: libcudacxx/include/cuda/__device/physical_device.h	94
2	1.862775	-1.862775	10.266514	8.403739	Processing Header File: libcudacxx/include/cuda/std/__cccl/prologue.h	548
3	1.327550	-1.327550	12.129615	10.802065	Processing Header File: libcudacxx/include/cuda/std/__iterator/concepts.h	436
4	1.137231	-1.137231	12.354520	11.217289	Processing Header File: libcudacxx/include/cuda/std/__type_traits/type_list.h	475
5	0.823850	-0.823850	3.764345	2.940495	Processing Header File: libcudacxx/include/cuda/std/__cccl/epilogue.h	548
6	0.823799	-0.823799	8.453583	7.629784	Processing Header File: cub/cub/util_type.cuh	278
7	0.617730	-0.617730	6.220260	5.602530	Processing Header File: libcudacxx/include/cuda/std/__iterator/iterator_traits.h	463
8	0.473442	-0.473442	6.549913	6.076471	Processing Header File: libcudacxx/include/cuda/std/__tuple_dir/vector_types.h	238
9	0.369159	-0.369159	3.947878	3.578719	Processing Header File: libcudacxx/include/cuda/std/__tuple_dir/tuple.h	243
10	0.346240	-0.346240	4.327040	3.980800	Processing Header File: libcudacxx/include/cuda/std/__utility/pair.h	403
11	0.339634	-0.339634	2.326101	1.986467	Processing Header File: libcudacxx/include/cuda/std/string_view	98
12	0.265380	-0.265380	3.289429	3.024049	Processing Header File: libcudacxx/include/cuda/std/__ranges/concepts.h	404
13	0.245630	-0.245630	1.505318	1.259688	Processing Header File: cub/cub/thread/thread_simd.cuh	137
14	0.224082	-0.224082	2.286644	2.062562	Processing Header File: libcudacxx/include/cuda/std/__functional/invoke.h	449
15	0.216362	-0.216362	2.873214	2.656852	Processing Header File: libcudacxx/include/cuda/std/__atomic/functions/cuda_ptx_generated.h	56