TensorRtSharp4.0

TensorRtSharp4.0 is a production-oriented TensorRT and CUDA bridge for .NET.

Current Status

Verified on 2026-06-12:

• TensorRT interface coverage: 0 missing rows.
• CUDA runtime interface coverage: 0 missing rows.
• Manifest inventory: 3271 API records across 102 manifests.
• Latest coverage report: artifacts/interface-coverage/interface-coverage-summary.md.
• Native validation: win-x64-trt11-cuda13-release configures and builds.
• Managed validation: solution build and project quality tests pass.

The project is now past raw interface coverage closure for the locally scanned headers. The active work is release hardening: sample coverage, documentation, package validation, and promotion of high-value deferred boundaries into safe public wrappers.

Start Here

English:

• Getting Started
• Installation Layout
• API Reference
• Windows API Completion
• API Coverage and Deferred Boundaries
• Sample Runners
• Runtime Package Strategy
• Runtime Distribution Strategy
• Local GitHub Actions Checks
• Package Consumer Validation
• Release Candidate Gate

Chinese articles:

• Getting Started
• Installation Layout
• API Reference
• Windows API Completion
• Latest Windows API Status
• Runtime Packages
• Runtime Distribution Strategy
• Local GitHub Actions Checks
• Package Consumer Validation
• Release Candidate Gate
• Sample Runners

Example And Smoke Path

Recommended common-example order:

1. MultiStream
2. DynamicShape
3. InferenceBindings
4. OnnxToEngine
5. Classification with your own classifier ONNX assets
6. YoloDet with your own YOLO-family ONNX assets

Recommended smoke order:

1. CudaSmokeRunner
2. TensorRtSmokeRunner
3. LifecycleSmokeRunner
4. OnnxToEngineSmokeRunner
5. NetworkBuilderSmokeRunner
6. Layer-specific network runners

See Sample Runners, samples/README.md, and smoke/README.md for commands and expected evidence lines.