Tuning policy cleanup 1/2

[bernhardmgruber]

• Proper designated init for ReducePolicy
• Extend all tests with the missing sub policy tests
• Rename policy tests
• Test operator<<

Fixes part of: #9378

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

📝 Walkthrough

Walkthrough

Device policy Catch2 tests were renamed to a properties naming scheme, and several tests now verify operator<< output while composite policy tests also compare named sub-policies built through aggregate and designated initialization paths.

Changes

Policy property test updates

Layer / File(s)	Summary
Simple test name and stream checks cub/test/catch2_test_device_adjacent_difference_env.cu, cub/test/catch2_test_device_find_bound_sorted_values_env.cu, cub/test/catch2_test_device_find_env.cu, cub/test/catch2_test_device_for_env.cu, cub/test/catch2_test_device_histogram_env.cu, cub/test/catch2_test_device_merge_env.cu, cub/test/catch2_test_device_merge_sort_env.cu, cub/test/catch2_test_device_run_length_encode_env.cu, cub/test/catch2_test_device_scan_by_key_env.cu, cub/test/catch2_test_device_select_env.cu, cub/test/catch2_test_device_segmented_radix_sort_env.cu	C2H_TEST labels were updated to the properties naming scheme, and these tests now assert non-empty operator<< output with std::ostringstream.
Nested policy construction and comparisons cub/test/catch2_test_device_memcpy_env.cu, cub/test/catch2_test_device_transform_env.cu, cub/test/catch2_test_device_partition_env.cu, cub/test/catch2_test_device_reduce_env.cu, cub/test/catch2_test_device_scan_env.cu, cub/test/catch2_test_device_segmented_scan_env.cu, cub/test/catch2_test_device_segmented_sort_keys_env.cu, cub/test/catch2_test_device_segmented_reduce_env.cu, cub/test/catch2_test_device_radix_sort_env.cu	Composite policy tests now build named sub-policies, compare aggregate and designated initialization paths, and add component-level equality and stream-output checks.

Possibly related PRs

• NVIDIA/cccl#8863: Updates the same ReducePassPolicy and ReducePolicy property tests in cub/test/catch2_test_device_reduce_env.cu.
• NVIDIA/cccl#9685: Adds the segmented radix sort policy type and its operator<<, matching the updated segmented radix sort test here.
• NVIDIA/cccl#9686: Updates the segmented reduce policy types and their streaming support, which this PR exercises in tests.

Suggested reviewers: shwina, NaderAlAwar, srinivasyadav18

---

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

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

cub/test/catch2_test_device_segmented_sort_keys_env.cu (1)

613-663: 🎯 Functional Correctness | 🔵 Trivial | ⚡ Quick win

suggestion: no STATIC_REQUIRE(p1 == p2) (or partitioning_threshold-specific check) is added alongside the new per-sub-policy comparisons. The component checks don't verify the top-level SegmentedSortPolicy aggregate equality/inequality operator itself.

diff

   STATIC_REQUIRE(p1_medium == p2_medium);
   STATIC_REQUIRE_FALSE(p1_medium != p2_medium);
+
+  STATIC_REQUIRE(p1 == p2);
+  STATIC_REQUIRE_FALSE(p1 != p2);

Source: Path instructions

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 7733a74a-f8a2-4126-960e-d8ac82405dd5

📥 Commits

Reviewing files that changed from the base of the PR and between a60fbe3 and 20a2890.

📒 Files selected for processing (20)

• cub/test/catch2_test_device_adjacent_difference_env.cu
• cub/test/catch2_test_device_find_bound_sorted_values_env.cu
• cub/test/catch2_test_device_find_env.cu
• cub/test/catch2_test_device_for_env.cu
• cub/test/catch2_test_device_histogram_env.cu
• cub/test/catch2_test_device_memcpy_env.cu
• cub/test/catch2_test_device_merge_env.cu
• cub/test/catch2_test_device_merge_sort_env.cu
• cub/test/catch2_test_device_partition_env.cu
• cub/test/catch2_test_device_radix_sort_env.cu
• cub/test/catch2_test_device_reduce_env.cu
• cub/test/catch2_test_device_run_length_encode_env.cu
• cub/test/catch2_test_device_scan_by_key_env.cu
• cub/test/catch2_test_device_scan_env.cu
• cub/test/catch2_test_device_segmented_radix_sort_env.cu
• cub/test/catch2_test_device_segmented_reduce_env.cu
• cub/test/catch2_test_device_segmented_scan_env.cu
• cub/test/catch2_test_device_segmented_sort_keys_env.cu
• cub/test/catch2_test_device_select_env.cu
• cub/test/catch2_test_device_transform_env.cu

[coderabbitai]

🧹 Nitpick comments (1)

cub/test/catch2_test_device_segmented_reduce_env.cu (1)

648-650: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

suggestion: unscoped enumerators (cub::BLOCK_REDUCE_WARP_REDUCTIONS, cub::LOAD_LDG) here vs. scoped cub::BlockReduceAlgorithm::.../cub::CacheLoadModifier::... in the reduce test. Harmless but inconsistent within the same cohort.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 1e01a70b-781d-4707-9548-e3ed2a081365

📥 Commits

Reviewing files that changed from the base of the PR and between 20a2890 and 56b9822.

📒 Files selected for processing (3)

• cub/test/catch2_test_device_reduce_env.cu
• cub/test/catch2_test_device_segmented_radix_sort_env.cu
• cub/test/catch2_test_device_segmented_reduce_env.cu

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

• cub/test/catch2_test_device_segmented_radix_sort_env.cu

[coderabbitai]

🧹 Nitpick comments (1)

cub/test/catch2_test_device_adjacent_difference_env.cu (1)

275-282: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

suggestion: this to_string lambda + REQUIRE(!to_string(p1).empty()) pattern is duplicated verbatim across ~9 policy test files in this PR stack. Consider hoisting it into a shared test header (e.g. a require_streamable_nonempty(p) helper) to avoid maintaining N copies.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 33425755-662a-4616-af59-05307a84af03

📥 Commits

Reviewing files that changed from the base of the PR and between 56b9822 and 88011d7.

📒 Files selected for processing (20)

• cub/test/catch2_test_device_adjacent_difference_env.cu
• cub/test/catch2_test_device_find_bound_sorted_values_env.cu
• cub/test/catch2_test_device_find_env.cu
• cub/test/catch2_test_device_for_env.cu
• cub/test/catch2_test_device_histogram_env.cu
• cub/test/catch2_test_device_memcpy_env.cu
• cub/test/catch2_test_device_merge_env.cu
• cub/test/catch2_test_device_merge_sort_env.cu
• cub/test/catch2_test_device_partition_env.cu
• cub/test/catch2_test_device_radix_sort_env.cu
• cub/test/catch2_test_device_reduce_env.cu
• cub/test/catch2_test_device_run_length_encode_env.cu
• cub/test/catch2_test_device_scan_by_key_env.cu
• cub/test/catch2_test_device_scan_env.cu
• cub/test/catch2_test_device_segmented_radix_sort_env.cu
• cub/test/catch2_test_device_segmented_reduce_env.cu
• cub/test/catch2_test_device_segmented_scan_env.cu
• cub/test/catch2_test_device_segmented_sort_keys_env.cu
• cub/test/catch2_test_device_select_env.cu
• cub/test/catch2_test_device_transform_env.cu

🚧 Files skipped from review as they are similar to previous changes (11)

• cub/test/catch2_test_device_find_env.cu
• cub/test/catch2_test_device_merge_env.cu
• cub/test/catch2_test_device_memcpy_env.cu
• cub/test/catch2_test_device_segmented_reduce_env.cu
• cub/test/catch2_test_device_segmented_sort_keys_env.cu
• cub/test/catch2_test_device_transform_env.cu
• cub/test/catch2_test_device_select_env.cu
• cub/test/catch2_test_device_segmented_scan_env.cu
• cub/test/catch2_test_device_partition_env.cu
• cub/test/catch2_test_device_scan_env.cu
• cub/test/catch2_test_device_radix_sort_env.cu

[github-actions]

🥳 CI Workflow Results

🟩 Finished in 9h 17m: Pass: 100%/242 | Total: 1d 08h | Max: 1h 08m | Hits: 99%/149992

See results here.