When to Use Hybridizer

Great fit when:

• Data-parallel workloads (SIMD/SIMT), large arrays, linear algebra, image/signal processing.
• Hot paths in C# or Java that dominate runtime.
• Need cross-platform performance without rewriting in CUDA/C++.

Considerations:

• Memory transfer cost GPU↔CPU.
• Algorithm parallelizability; control flow divergence.
• Interop with existing native libraries.
• Known limitations (see Reference → Glossary/Limitations).