This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS570LS3137: Review Compiler and linker settings

Part Number: TMS570LS3137

Please support by reviewing the compiler and linker settings used in our project and suggest any complier options which would benefit in improving the performance of the CPU

Compiler and Compiler Options

Texas Instruments in version 15.12.4.LTS.

-mv=7R4 

-c 

--endian=big 

-k 

-g 

--gcc 

--float_support=vfpv3d16 

--abi=eabi 

-mt 

--opt_level=3 

--opt_for_speed=5 

--opt_for_cache 

--unaligned_access=off 

--align_structs=4

 

Linker and Linker Options

Texas Instruments in version 15.12.4.LTS.

-mv7r4 

-a 

--rom_model 

--warn_sections 

--reread_libs 

--unused_section_elimination=on 

--cinit_compression=rle 

--compress_dwarf=on 

--copy_compression=rle 

--reread_libs 

--display_error_number 

-m StartApplication.map 

--be32 

--diag_suppress=10068

  • Prathibha Rajaram said:
    --endian=big 

    This is a case where you have to match how your hardware is configured.  I presume all the options which are closely related to the hardware configuration have been chosen correctly.

    Prathibha Rajaram said:
    -mt 

    The equivalent long form option is --code_state=16.  And the opposite is --code_state=32.  The best choice for this option is found by experiment.  Mixing files built with different states is supported.  It may make sense to compile compute intensive code with --code_state=32, and other code with --code_state=16.  You can do the same thing on a function by function basis with #pragma CODE_STATE.

    Prathibha Rajaram said:

    --opt_level=3 

    --opt_for_speed=5 

    The best settings for these options can be found by using a feature of Code Composer Studio called Optimizer Assistant.

    Prathibha Rajaram said:
    --opt_for_cache 

    The following quote comes from the article Compiler Option Cleanup.

    The –opt_for_cache option was an experiment that turned out not to work well in many cases. We advise that you try different –opt_for_speed settings to get the best performance on cache devices.

    Prathibha Rajaram said:
    --unaligned_access=off 

    Most Cortex-R4 systems are configured to supported unaligned access.  But if your system does not support unaligned access, then you have to set this option to off.

    Thanks and regards,

    -George