[uurl]

Summary

• Add f32 and f64 mappings for libm::modf{f} and libm::remquo{f} to the generated libdevice math shard.
• Reuse the frexp out-parameter call helper and pack the results according to the Rust tuple layout. Normalize the remquo quotient to zero when the remainder is NaN.
• Extend CUDA 13.0 ABI evidence, generator and lowering tests, and add the math_out_params GPU example.

The shard now covers 108 of the 124 public f32/f64 exports in libm 0.2.16.

Validation

• cargo oxide fmt --check
• cargo run -p cuda-intrinsics-gen --locked -- check
• cargo test -p cuda-intrinsics-gen -p dialect-mir -p mir-importer -p mir-lower --all-targets --locked
• cargo clippy -p cuda-intrinsics-gen -p dialect-mir -p mir-importer -p mir-lower --all-targets --locked -- -D warnings
• cargo oxide run math_out_params: 180 GPU checks passed across f32 and f64
• cargo oxide run math_special: 776 GPU checks passed
• just check

Depends on #1650. Refs #1224.

This completes the planned out-parameter mappings. The remaining 16 libm exports require separate semantic handling; this PR does not close #1224.

Testing

• just -f cuda-oxide/Justfile check passes (the local mirror of CI: fmt, clippy, tests, guards, docs)
• cargo oxide run <example> passes, or cuda-oxide/scripts/smoketest.sh -o '^<example>$'
• New example added (if applicable)

Checklist

• All commits signed off (git commit -s)
• SPDX headers on new source files

[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.