Model Configuration Parameters: Code Generation Interface

The Code Generation > Interface category includes parameters for configuring the interface of the generated code. These parameters require a Simulink^® Coder™ license. Additional parameters available with an ERT-based target require an Embedded Coder^® license. Generating code for deep learning models using NVIDIA^® CUDA^® deep neural network library (cuDNN) or TensorRT™ high performance inference libraries for NVIDIA GPUs requires a GPU Coder™ license.

On the Configuration Parameters dialog box, the following configuration parameters are on the Code Generation > Interface pane.

Parameter	Description
Code replacement library	Specify a code replacement library the code generator uses when producing code for a model.
Code replacement libraries	Specify multiple code replacement libraries the code generator use when producing code for a model.
Shared code placement	Specify the location for generating utility functions, exported data type definitions, and declarations of exported data with custom storage class.
Support: floating-point numbers	Specify whether to generate floating-point data and operations.
Support: non-finite numbers	Specify whether to generate non-finite data and operations on non-finite data.
Support: complex numbers	Specify whether to generate complex data and operations.
Support: absolute time	Specify whether to generate and maintain integer counters for absolute and elapsed time values.
Support: continuous time	Specify whether to generate code for blocks that use continuous time. This parameter does not apply when you use Embedded Coder for a component model that is configured with a service interface.
Support: variable-size signals	Specify whether to generate code for models that use variable-size signals.
Code interface packaging	Select the packaging for the generated C or C++ code interface.
Multi-instance code error diagnostic	Select the severity level for diagnostics displayed when a model violates requirements for generating multi-instance code.
Pass root-level I/O as	Control how root-level model input and output are passed to the reusable `model_step` function.
Remove error status field in real-time model data structure	Specify whether to log or monitor error status.
Include model types in model class	Specify whether to generate model type definitions in a model class.
Array layout	Specify layout of array data for code generation as column-major or row-major
External functions compatibility for row-major code generation	Select diagnostic action if Simulink encounters a function that has no specified array layout
Generate C API for: signals	Generate C API data interface code with a signals structure.
Generate C API for: parameters	Generate C API data interface code with parameter tuning structures.
Generate C API for: states	Generate C API data interface code with a states structure.
Generate C API for: root-level I/O	Generate C API data interface code with a root-level I/O structure.
ASAP2 interface	Generate code for the ASAP2 data interface.
External mode	Generate code for the external mode data interface. When using Embedded Coder for a component model that is configured with a service interface, this parameter is not relevant and, therefore, is not supported.
Transport layer	Specify the transport protocol for communications.
MEX-file arguments	Specify arguments to pass to an external mode interface MEX-file for communicating with executing targets.
Static memory allocation	Control memory buffer for external mode communication.
Static memory buffer size	Specify the memory buffer size for external mode communication.
Automatically allocate static memory	Automatically allocate static memory for buffers used in external mode communication.
Maximum duration	Specify, in base rate steps, maximum duration that software must consider when determining size of static memory required for external mode communication.
Log signals to MDF file	Log signal and output data from generated code to MDF file.
Target library	Specify the target deep learning library used during code generation. `cuDNN` or `TensorRT` requires a GPU Coder license.
Learnables Compression	Specify the compression type, `"none"` or `"bfloat16"`.
ARM Compute Library version	Specify the version of ARM^® Compute Library.
ARM Compute Library architecture	Specify the ARM architecture supported in the target hardware.
Deep Learning Data Type	Specify the data type to be used by TensorRT library for deep learning code generation. This parameter requires a GPU Coder license.
Auto tuning	Use auto tuning for cuDNN library. Enabling auto tuning allows the cuDNN library to find the fastest convolution algorithms. This parameter requires a GPU Coder license.

These configuration parameters are under the Advanced parameters.

Parameter	Description
Support non-inlined S-functions	Specify whether to generate code for non-inlined S-functions. This parameter does not apply when you use Embedded Coder for a component model that is configured with a service interface.
Maximum word length	Specify a maximum word length, in bits, for which the code generation process generates system-defined multiword type definitions.
Buffer size of dynamically-sized string (bytes)	Number of bytes of the character buffer generated for dynamic string signals without maximum length.
Multiword type definitions	Specify whether to use system-defined or user-defined type definitions for multiword data types in generated code.
Use dynamic memory allocation for model initialization	Control how the generated code allocates memory for model data.
Single output/update function	Specify whether to generate the `model_step` function. This parameter does not apply when you use Embedded Coder for a component model that is configured with a service interface.
Terminate function required	Specify whether to generate the `model_terminate` function.
Combine signal/state structures	Specify whether to combine global block signals and global state data into one data structure in the generated code You cannot clear this parameter when you use Embedded Coder for a component model that is configured with a service interface.
Generate separate internal data per entry-point function	Generate a model's block signals (block I/O) and discrete states (DWork) acting at the same rate into the same data structure.
MAT-file logging	Specify MAT-file logging. This parameter does not apply when you use Embedded Coder for a component model that is configured with a service interface.
MAT-file variable name modifier	Select the text to add to MAT-file variable names.
Existing shared code	Specify folder that contains existing shared code
Suppress generation of fixed-point word size checks	Control generation of preprocessor directives that perform fixed-point word size checks
Remove disable function	Remove unreachable (dead-code) instances of the `disable` functions from the generated code for ERT-based systems that include model referencing hierarchies.
Remove reset function	Remove unreachable (dead-code) instances of the `reset` functions from the generated code for ERT-based systems that include model referencing hierarchies.
LUT object struct order for even spacing specification	Change the order of the fields in the generated structure for a lookup table object whose specification parameter is set to even spacing.
LUT object struct order for explicit value specification	Change the order of the fields in the generated structure for a lookup table object whose specification parameter is set to explicit value.
Generate destructor	Specify whether to generate a destructor for the C++ model class.
Use dynamic memory allocation for model block instantiation	Specify whether generated code uses the operator `new`, during model object registration, to instantiate objects for referenced models configured with a C++ class interface.
Ignore custom storage classes	Specify whether to apply or ignore custom storage classes.
Ignore test point signals	Specify allocation of memory buffers for test points.
Implement each data store block as a unique access point	Create unique variables for each read/write operation of a Data Store Memory block.
Generate full file banner	Generate full header banner including time stamp.

The following parameters under the Advanced parameters are infrequently used and have no other documentation.

Parameter	Description
`GenerateSharedConstants`	Control whether the code generator generates code with shared constants and shared functions. Default is `on`. When turned `off`, the code generator does not generate shared constants.
`InferredTypesCompatibility`	For compatibility with legacy code including `tmwtypes.h`, specify that the code generator creates a preprocessor directive `#define __TMWTYPES__` inside `model.h`.
`TargetLibSuffix` `character vector` - `''`	Control the suffix used for naming a target's dependent libraries (for example, `_target.lib` or `_target.a`). If specified, the character vector must include a period (.). (For generated model reference libraries, the library suffix defaults to `_rtwlib.lib` on Windows^® systems and `_rtwlib.a` on UNIX^® systems.). This parameter does not apply for model builds that use the toolchain approach, see Library Control Parameters
`TargetPreCompLibLocation` `character vector` - `''`	Control the location of precompiled libraries. If you do not set this parameter, the code generator uses the location specified in `rtwmakecfg.m`.
`IsERTTarget`	Indicates whether or not the currently selected target is derived from the ERT target.
`CPPClassGenCompliant`	Indicates whether the target supports the ability to generate and configure C++ class interfaces to model code.
`ConcurrentExecutionCompliant`	Indicates whether the target supports concurrent execution
`UseToolchainInfoCompliant`	Indicate a custom target is toolchain-compliant.
`ModelStepFunctionPrototypeControlCompliant`	Indicates whether the target supports the ability to control the function prototypes of initialize and step functions that are generated for a Simulink model. This parameter does not apply when you use Embedded Coder for a component model that is configured with a service interface.
`ParMdlRefBuildCompliant`	Indicates if the model is configured for parallel builds when building a model that includes referenced models.
`CompOptLevelCompliant` `off`, `on`	Set in `SelectCallback` for a target to indicate whether the target supports the ability to use the Compiler optimization level parameter to control the compiler optimization level for building generated code. Default is `off` for custom targets and `on` for targets provided with the Simulink Coder and Embedded Coder products.
`ModelReferenceCompliant` character vector - `off`, `on`	Set in `SelectCallback` for a target to indicate whether the target supports model reference.

The following parameters are for MathWorks use only.

Parameter	Description
`ExtModeTesting`	For MathWorks use only.
`ExtModeIntrfLevel`	For MathWorks use only.
`ExtModeMexFile`	For MathWorks use only.

Model Configuration Parameters: Code Generation Interface

Related Topics