Due to the U.S. Thanksgiving holiday, please expect delayed responses during the week of 11/22.

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.

CCS/HERCULES-DSPLIB: Compiling Hercules library for AM65x R5F


Tool/software: Code Composer Studio

Since the latest RTOS SDK (6.02) which would have the DSPLib incorporated in the SDK is not out, we are trying to build the Hercules-Lib for the cortex R5F on the AM6546 SoC that can be incorporated in the TI-RTOS SDK on our Linux system. We followed the instructions at:


But this generates a .lib (windows compatible) library. We changed the tool-chain from TI v18.12.4 LTS to GNU tool-chain (Linaro) which creates the .a linux static library file. However, this failed with the following build log.


Adding the location of arm_math.h to the include path led to a bunch of other errors. So I was wondering if this experiment has laready been conducted and if there is a solution to it.

  • Hi,

    I see the Hercules DSP library installer is only provided as a Windows installer. It should be possible to move the installed content to a Linux system if you care to build it there.

    >> But this generates a .lib (windows compatible) library.

    The generated .lib is for an R5F target, and isn't specific to Windows. The .lib should link with code compiled on a Linux system using TI v18.12.4 LTS.

    >> GNU tool-chain (Linaro) which creates the .a linux static library file.

    No, the GNU tool chain has never been tried with the Hercules DSP library for AM65xx R5F. However, I would expect this to work (perhaps some modification is required for header / make files for the particular GNU tool chain version) since Hercules DSP library is simply an older version of the CMSIS library, and the CMSIS DSP library typically has good built-in support for the GNU ARM tool chain.


  • Thanks Frank. We were able to incorporate the .lib and perform convolution in our code.

    Also we needed to provide floating point support option (VFPv3D16) to allow floating point generation. Linking fails otherwise.