Advanced Topics

This section covers advanced TensorRT features and configuration options. For Version Compatibility and hardware compatibility, use the top-level Inference Library entry (promoted from this hub in 11.2.1).

• Refitting an Engine
  • Weight-Stripping
  • Refitting a Weight-Stripped Engine Directly from ONNX
  • Weight-Stripping Work with Lean Runtime
  • Fine-Grained Refit Build
  • Stripping Weights with Fine-Grained Refit Build
• Precision Control
  • Algorithm Selection and Reproducible Builds
  • Strongly Typed Networks
  • Control of Computational Precision
• Data Formats and Tensors
  • I/O Formats
  • Sparsity
  • Empty Tensors
  • Reusing Input Buffers
• Data Format Descriptions
• Engine Tools and Debugging
  • Engine Inspector
  • Optimizer Callbacks
  • Preview Features
  • Debug Tensors
• Weight Streaming
• Multi-Device Inference
  • Setup
  • NCCL Version Notes
  • Platform Support
  • Feature Compatibility