Deep Learning Data Type

Data type for generating code that uses NVIDIA TensorRT high performance inference library

Since R2024a

Model Configuration Pane: Code Generation / Interface

Description

The Data type parameter specifies the data type to be used by TensorRT library for deep learning code generation.

Dependencies

• This parameter requires a GPU Coder™ license.

• To set the data type for TensorRT libraries, in the Configuration Parameters dialog, in the Code Generation settings, set Language to C++ and enable Generate GPU Code. Then in Code Generation > Interface settings, in the Deep learning section, set Target library to TensorRT.

Settings

"fp32" (default) | "fp16"

fp32

Generate code in 32-bit floating-point.

fp16

Generate code in half-precision floating-point.

Recommended Settings

Application	Setting
Debugging
Traceability
Efficiency
Safety precaution

Programmatic Use

Parameter: DLDataType

Value: "fp32" | "fp16"

Default: "fp32"

Version History

Introduced in R2024a