CUDA: add auto constant tests\n\nAdds kernel-based tests for device-s…

[CodersAcademy006]

This PR adds kernel-based tests for device-side compile-time constants in Numba-CUDA:

• Integer, float, and boolean literals

• Constant expressions (e.g., 2 + 3, 4 * 5)

• Mixed constant arithmetic

• Constant indexing

Tests are skipped under cudasim due to divergent semantics, and NumPy is used as the reference oracle. The scope is intentionally limited to safe, well-defined CUDA constant lowering (no globals, closures, or runtime values).

This continues the systematic porting of CPU-side tests to CUDA, directly contributing to issue #515.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[greptile-apps]

Greptile Summary

Added comprehensive test coverage for CUDA device-side compile-time constants, including integer/float/boolean literals, constant expressions, mixed arithmetic, and constant indexing. Tests are properly skipped under cudasim where semantics differ, use NumPy as reference oracle, and follow repository conventions with correct SPDX header and test structure.

• Integer constant test (42)
• Float constant test (3.5)
• Boolean constant test (True)
• Arithmetic expression test (2 + 3 * 4 = 14)
• Mixed type arithmetic test (1.5 + 2 = 3.5)
• Constant indexing test (inp[2])

The implementation follows established patterns from other CUDA tests in the repository, properly uses CUDATestCase, includes appropriate skip_on_cudasim decorator, and provides a reusable _launch_1d helper method. This directly contributes to issue #515 by porting CPU-side auto-constant tests to CUDA.

Confidence Score: 5/5

• This PR is safe to merge with no identified issues
• The test file is well-structured, follows all repository conventions including proper SPDX header, uses appropriate testing patterns (CUDATestCase, skip_on_cudasim), and provides comprehensive coverage of compile-time constant scenarios. The arithmetic is correct, tests are isolated and independent, and the implementation aligns with the stated goal of porting CPU tests to CUDA.
• No files require special attention

Important Files Changed

Filename	Overview
numba_cuda/numba/cuda/tests/test_auto_constants.py	New test suite for compile-time constants with proper structure, correct SPDX header, and comprehensive coverage of integer, float, boolean, arithmetic, and indexing scenarios

[greptile-apps]

_{1 file reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}