- Use Simulink Data Dictionary (SLDD) to Segment Variables: In the Simulink Data Dictionary, group related variables into smaller "Simulink.Bus" objects. Assign the "Simulink.Bus" objects to subsystems/functions within the model.
- Custom Storage Classes: Define custom storage classes for the variables using Embedded Coder. Group variables by assigning them to specific "Simulink.Signal" or "Simulink.Parameter" objects with unique storage classes.
- Manual Splitting in the Generated Code: Post-process the generated code to reorganize the "ARID_DEF" struct into smaller structs. Use "coder.cstructname" to control how specific variables appear in the struct.
- Subsystem Partitioning: Partition your Simulink model into smaller subsystems that encapsulate related variables. The generated code will create structs local to each subsystem, splitting the large "ARID_DEF" struct.
Unrolling ARID_DEF structs
61 views (last 30 days)
Show older comments
Hi!
Generating code for a .slx component gives a very large ARID_DEF struct which contains all of the variables in the component, in all of the files and functions. This makes the struct very large and hard to integrate/optimize onto hardware with limited memory.
It would be beneficial to "unroll" this struct and divide it into smaller structs, such that critical variables can be put on fast memory and non-critical variables onto slower memory. Right now I have to decide where to place the entire struct.
How can one achieve this?
For example:
function1.m
variableA
variableB
function2.m
variableC
variableD
shall not result in
typedef struct
{
variableA;
variableB;
variableC;
variableD;
}
ARID_DEF_COMPONENTNAME_T;
but rather
typedef sub_struct1
{
variableA;
variableB;
}
ARID_DEF_SUBSTRUCT1_T;
typedef sub_struct2
{
variableC;
variableD;
}
ARID_DEF_SUBSTRUCT2_T;
0 Comments
Answers (1)
Maneet Kaur Bagga
on 20 Nov 2024 at 10:49
Hi,
As per my understanding, to optimize the large "ARID_DEF" struct by splitting it into smaller sub-structs, you can use the following as a possible workaround:
Please refer to the following MathWorks Documentation for "coder.cstructname" for better understanding:
Hope this helps!
0 Comments
See Also
Categories
Find more on Multicore Processor Targets in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!