cuda::std::simd: Add shr, shl, bit_reverse

[fbusato]

Description

Add C++29 binding to cuda::std::simd

Fixes #10041

[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: 833c4297-dfc1-4187-9d2b-f8df32c92494

📥 Commits

Reviewing files that changed from the base of the PR and between db424ab and 8b45edf.

📒 Files selected for processing (4)

• libcudacxx/include/cuda/std/__simd/bit/scalar.h
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/bit_reverse.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shl.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shr.pass.cpp

🚧 Files skipped from review as they are similar to previous changes (4)

• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shr.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/bit_reverse.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shl.pass.cpp
• libcudacxx/include/cuda/std/__simd/bit/scalar.h

---

📝 Walkthrough

Summary by CodeRabbit

• New Features
  • Added scalar SIMD bit-manipulation APIs for fixed-size vector lanes, including bit reversal, byte swapping, bit ceil/floor, single-bit checks, shift-left/right (vector and scalar counts), rotations (vector counts), and bit-counting queries.
• Tests
  • Added new pass tests for bit_reverse, shl, and shr, verifying per-lane correctness, return types, noexcept behavior, and overload/type constraints.

Walkthrough

Changes

The SIMD bit header now delegates to a scalar implementation header. The implementation adds per-lane bit manipulation, shift, rotate, and bit-counting APIs with constraints. Tests cover bit_reverse, shl, and shr values, types, noexcept, and overload availability.

SIMD bit operations

Layer / File(s)	Summary
Scalar generators and dependencies libcudacxx/include/cuda/std/__simd/bit.h, libcudacxx/include/cuda/std/__simd/bit/scalar.h	Adds per-lane generators and replaces the aggregate header’s prior dependency list with __simd/bit/scalar.h.
Public SIMD bit APIs libcudacxx/include/cuda/std/__simd/bit/scalar.h	Exposes constrained wrappers for bit transforms, shifts, rotates, and bit-counting operations.
Bit-reverse validation libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/bit_reverse.pass.cpp	Tests bit-reversal results, return types, noexcept, and supported element types.
Shift API validation libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shl.pass.cpp, libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shr.pass.cpp	Tests vector and scalar shifts, boundary counts, elementwise results, and overload constraints.

Assessment against linked issues

Objective	Addressed	Explanation
Implement the SIMD portion of P3793R2 [#10041]	✅

Possibly related issues

• [FEA]: C++29 <cuda/std/bit> Implement Bit permutations #10056 — Directly related to the SIMD bit_reverse binding; this PR covers bit_reverse and additional SIMD bit operations.

Possibly related PRs

• NVIDIA/cccl#9993 — Provides the scalar shl/shr helpers used by the new SIMD implementations.
• NVIDIA/cccl#10004 — Covers the scalar rotate helpers delegated to by SIMD rotl/rotr.

Suggested reviewers: davebayer, jacobfaib

---

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

[coderabbitai]

Actionable comments posted: 2

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 01eaa830-7e69-42c3-8435-6b23a646c7ab

📥 Commits

Reviewing files that changed from the base of the PR and between d7823c5 and db424ab.

📒 Files selected for processing (5)

• libcudacxx/include/cuda/std/__simd/bit.h
• libcudacxx/include/cuda/std/__simd/bit/scalar.h
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/bit_reverse.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shl.pass.cpp
• libcudacxx/test/libcudacxx/std/numerics/simd/simd.bit/shr.pass.cpp

[github-actions]

🥳 CI Workflow Results

🟩 Finished in 2h 31m: Pass: 100%/120 | Total: 1d 10h | Max: 1h 17m | Hits: 89%/427983

See results here.