Learn more about Real-Time Workshop Embedded Coder

Setting Up a Model to Generate Code for Host Simulations and Target Deployment

On this page…
Compiling Generated Code That Supports Portable Word Sizes Portable Word Sizes Limitation

When configuring a model for SIL simulation with the Real-Time Workshop product and when processor word sizes differ between host and target platforms, you can configure the model to use an emulation hardware option. That option guarantees bit-true agreement for integer and fixed-point operations between host system simulation results and target deployment.

In addition to the emulation hardware option, the Real-Time Workshop Embedded Coder product provides an option for configuring a model to use portable word sizes. Using this second option, you can configure a model to generate code that you can compile without changing it for a platform. You can then use the code for SIL simulation on a host system and for deployment on a target system.

Note   When you use either of the above methods for SIL simulation, differences between your host and target processor characteristics can cause differences in the simulation results of the generated ERT S-function wrapper block when compared to the original model or subsystem. If processor word sizes differ between host and target platforms, and you use neither of the above methods for SIL simulation, there are likely to be differences between host simulation results and target execution results. When using portable word sizes for SIL simulation, subtle differences in host and target processor behavior can still cause host simulation results to differ from target execution results, but this is not as common. For more information, see Portable Word Sizes Limitation.

To configure a model to use portable word sizes, set model configuration parameters.

Set...	To...
Hardware Implementation > Emulation hardware > None	Selected
Real-Time Workshop > Interface > Create Simulink (S-Function) block	Selected
Real-Time Workshop > Interface > Enable portable word sizes	Selected

When you generate code for a model with the preceding parameter settings, the code generator conditionalizes data type definitions:

• tmwtypes.h supports SIL simulation on the host system

• Real-Time Workshop types support deployment on the target system

For example, in the following generated code, the first two lines define types for SIL simulation on a host system. The bold lines define types for target deployment.

#ifdef PORTABLE_WORDSIZES    /* PORTABLE_WORDSIZES defined */
# include "tmwtypes.h"
#else                        /* PORTABLE_WORDSIZES not defined */
#define __TMWTYPES__
#include <limits.h>
...
typedef signed char int8_T;
typedef unsigned char uint8_T;
typedef int int16_T;
typedef unsigned int uint16_T;
typedef long int32_T;
typedef unsigned long uint32_T;
typedef float real32_T;
typedef double real64_T;
...
#endif                       /* PORTABLE_WORDSIZES */

For an example of how to configure a model to maintain bit-true agreement between host simulation and target deployment, and generate code that is portable between the host and target systems, see rtwdemo_sil_pil.

Compiling Generated Code That Supports Portable Word Sizes

When compiling generated code that supports portable word sizes for SIL testing, you need to pass the definition PORTABLE_WORDSIZES to the compiler.

For example:

-DPORTABLE_WORDSIZES

To build the same code for target deployment, compile the code without the PORTABLE_WORDSIZES definition.

Back to Top

Portable Word Sizes Limitation

When performing SIL testing, using the Enable portable word sizes model configuration parameter, numerical results of the S-function simulation on the MATLAB host might differ from results on the actual target. This difference is due to differences in target characteristics, such as:

• C integral promotion in expressions may be different on the target processor

• Signed integer division rounding behavior may be different on the target processor

• Signed integer arithmetic shift right may behave differently on the target processor

• Floating-point precision may be different on the target processor

Back to Top

About Software-In-the-Loop Simulation Extensions

Generating an S-Function Wrapper for SIL Testing

© 1984-2009- The MathWorks, Inc.    -   Site Help   -   Patents   -   Trademarks   -   Privacy Policy   -   Preventing Piracy   -   RSS