Model Build Behavior with audesse-blocks 2.1 vs 2.0

[Tom Stallard]

#21300

I uninstalled audesse-blocks_2.0 and installed audesse-blocks_2.1. For the same model, I observed the following:
– Receiving the following error:
c:/users/it/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: section .non_cacheable_data LMA [00461008,0046100f] overlaps section .sram_data LMA [00461008,004612cf]
c:/users/it/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: region `int_sram_results’ overflowed by 256 bytes

– When the model is created in a different folder, the build process still searches for the model folder inside the toolbox installation directory. This effectively forces the project to be created within the audesse_blocks folder.
– If the audesse blocks are uninstalled, the model builds successfully without errors, provided the linker file from audesse-blocks_2.1 is copied into the S32K3 toolbox installation folder.

[Connor]

#21302

Hi Tom,

We will attempt to replicate the path bug you’re seeing and get back to you ASAP.

In the meantime, for your RAM overflow:
– Check the generated code for any moving average blocks, generally these eat up RAM and there are other ways to achieve filtering.
– Let us know which specific sample model from audesse_blocks-2.1 that you pasted your application code into.
– Confirm that you’re on NXP MBDT 1.7 and that previous versions are uninstalled
– Let us know which MATLAB version you’re running

• This reply was modified 3 months, 1 week ago by Connor.

[Tom Stallard]

#21324

Hi Connor,

– Will check for moving average blocks and update
– FreeMasteroverCAN
– We are on MBDT 1.7 and previous versions were uninstalled
– MATLAB 2024a will be shifting to 2025a soon

[Connor]

#21329

Hi Tom,

I was not able to replicate your build errors in either 2022a or 2025b. If your model is based on the FreeMaster sample, you should be able to move that model anywhere in your file system and have it automatically generate the config and build files. Some extra troubleshooting things to try:
– Make sure your MATLAB addons path does not have spaces (occasionally causes stability problems)
– Make sure the Embedded Coder Support Package for ARM Cortex-M Processors addon is installed
– Review the installation steps in https://www.audesseinc.com/resources/flexcase-e/simulink-basics/ to make sure nothing got missed

If you are planning to migrate MATLAB versions soon, I recommend 2025b instead of 2025a since “b” versions tend to be more stable, and we are currently testing our models in 2025b (along with 2022a for compatibility)

Also, we will likely release Audesse blocks v2.2 this week to make the templates more portable, I will post here when released.

[Tom Stallard]

#21331

Hi Connor,

– Definitely the path does not have spaces
– Embedded Coder support package for Arm Cortex-M was already installed

I will wait for you to release the v2.2 blocks and give a shot at it again.

[Connor]

#21375

Hi Tom, v2.2 is now available at the downloads page

[Tom Stallard]

#21482

Hi Connor,

I have tried with Audesse Blocks v2.2 and getting the following error:

c:/users/tom/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: ../ASW.elf section .int_results' will not fit in regionint_sram_results’
c:/users/tom/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: Memory for standby ram overflow
c:/users/tom/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: Memory for standby ram overflow
c:/users/tom/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: section .non_cacheable_data LMA [00481008,0048100f] overlaps section .sram_data LMA [00481008,00486eab]
c:/users/tom/appdata/roaming/mathworks/matlab_add-ons/toolboxes/nxp_mbdtoolbox_s32k3/tools/build_tools/gcc_v10.2/gcc-10.2-arm32-eabi/bin/../lib/gcc/arm-none-eabi/10.2.0/../../../../arm-none-eabi/bin/real-ld.exe: region `int_sram_results’ overflowed by 256 bytes

Let me know if there are any further optimisation settings i should take care of in model settings to avoid this. Currently I am at Maximum (Balance RAM and Speed).

Thanks,
Tom

[Connor]

#21483

Hi Tom,

Just double checking the previous advice:

– Did you start with the FreeMaster_over_CAN_k3.slx template from v2.2 to paste your functions into? The templates inherit the recommended build optimizations from NXP
– Did you convert any moving averages to a RAM-friendly version?

If you confirm these aren’t the issue, I can continue digging

[Tom Stallard]

#21485

Hi Connor,

We managed to solve this issue. Turned out it worked in another laptop with MATLAB 2025. I was trying on MATLAB 2024a.

Thanks,
Tom

[Author]

Posts