Metadata-Version: 2.4
Name: starwave
Version: 7.1.0
Summary: Scalar, VRZ, VTI, elastic and GSLS propagation with native and PyTorch backends
Requires-Python: >=3.10,<3.13
Description-Content-Type: text/plain
Requires-Dist: torch>=2.5,<2.6
Requires-Dist: numpy>=1.23
License-File: native/LICENSE.SWEEP
License-File: cpu_validation/reference/LICENSE
License-File: starwave/_deepwave/LICENSE.Deepwave.txt
License-File: THIRD_PARTY_NOTICES.md
License-File: LICENSE.NVIDIA-CUDA
License-File: LICENSE.GCC-RUNTIME
License-File: BINARY_DISTRIBUTION_NOTICES.txt

StarWave 7.1.0

Binary-only Linux x86-64 package from separately frozen source 0.1.0.dev16.
Python 3.10–3.12, PyTorch >=2.5,<2.6, NumPy >=1.23 and glibc >=2.35.
Native CUDA Runtime 11.8.89 is static; SASS sm70/75/80/86/89/90 and compute80 PTX are included.
Users supply a compatible NVIDIA driver and CUDA-enabled Torch for GPU execution.
The complete public Python runtime includes explicit pure-Torch eager/compile full/checkpoint backends and GSLS.
Standalone single-SLS is removed; use the documented GSLS sls_compat parameterization.
Four native libraries cover core, elastic, GSLS CPU and GSLS CUDA. No automatic build or device fallback occurs.
Native host loading and CPU checks do not establish GPU numerical or multi-GPU acceptance.
See starwave/_binary/release.json for the exact source identity, integrity hashes and scoped validation status.
The project owner has not selected an overall StarWave license; retained third-party licenses apply to their respective components.
