Release notes
0.4.11
- First release of 0.4 branch on public NuGet.
0.4.10
- Update to CUDA 9.1
- CUDA SDK no longer required.
0.4.9
- Remove type constraints on Tensor.arange and Tensor.linspace.
0.4.8
- Support for dynamic loading of BLAS libraries for host tensors. Use Tensor.Host.Cfg.BLASLib to configure.
0.4.7
- Experimental support for macOS (without CUDA).
0.4.6
- Fix double-precision BLAS operations on CUDA GPU.
0.4.5
- No changes.
0.4.4
- Fix tensor_mkl.dll not being found.
- Update Tensor.Sample sample project.
- Add SourceLink support (experimental).
0.4.3
- Documentation improvements.
- Add Tensor.Sample sample project.
0.4.2
- Improve CUDA and CUBLAS initialization.
0.4.1
- First version to support .NET Standard 2.0.
- Optimized host backend.
- Redesigned API.