• This 1.2.0 version will no longer support old gcc 10 verison, and it now only support Nuclei Toolchain 2023.10(gcc13 and clang17) or later. The major changes that can be felt are as follows:

  • The prefix of toolchain has changed from riscv-nuclei-elf- to riscv64-unknown-elf-

  • The -march option has changed a lot, for example:

    • b extension changed to _zba_zbb_zbc_zbs extension,

    • p extension changed to _xxldsp or _xxldspn1x or _xxldspn2x or _xxldspn3x extensions which means stardard DSP extension, Nuclei N1, N2, N3 DSP extensions

    • v extension changed to v or _zve32f or _zve64f extensions according to the riscv cpu isa used

    These extensions need be combined in a certain order to get a correct arch name to match the prebuilt library name, please be cautious

  • The name of libraries changed due to -march in gcc13 updated, for example, the library named libnmsis_dsp_rv32imacb.a is now named libnmsis_dsp_rv32imac_zba_zbb_zbc_zbs.a since b extension changed to _zba_zbb_zbc_zbs

This is release 1.2.0 version of NMSIS-DSP library.

  • Defined NUCLEI_DSP_DEFAULT, NUCLEI_DSP_N1, NUCLEI_DSP_N2, NUCLEI_DSP_N3 in riscv_math_types.h according to gcc options. This means that if compile with --march=rv32imafc_xxldspn1x, the NUCLEI_DSP_N1 will defined, if compile with --march=rv32imafc_xxldspn2x, the NUCLEI_DSP_N1 and NUCLEI_DSP_N2 will defined, and so on

  • Optimize some functions with DSP N1/N2/N3 (such as FilteringFunctions, TransformFunctions, ComplexMathFunctions)

  • RVV intrinsic APIs is update to v0.12.0

  • Add f16 support(include f16 rvv extension support)

  • Fix the use of expd80 instruction(Nuclei default dsp instruction)

  • Fix some testcases bugs(such as MatrixFunctions, TransformFunctions)


This is release 1.1.1 version of NMSIS-DSP library.

  • Sync changes from CMSIS DSP commit 1d9e38a, version after v1.14.4

  • Optimized more for RVP/RVV

  • Add support for RV32 Vector

  • Some bugfix(riscv_mat_inverse_f32.c rvv fix, riscv_offset_q15.c p fix, riscv_fir_q15.c rvv fix etc.)


This is release 1.1.0 version of NMSIS-DSP library.

  • Sync changes from CMSIS 5.9.0 release

  • Optimized more for RVP/RVV

  • Add experimental support for RV32 Vector


This is release 1.0.3 version of NMSIS-DSP library.

  • Update build system for NMSIS-DSP library

  • Rename RISCV_VECTOR to RISCV_MATH_VECTOR in header file and source code

  • Using new python script to generate NMSIS-DSP library

  • Fix riscv_float_to_q31 function for rv64imafcv target

  • Change vfredsum to vfredusum when using vector intrinsic function due to vector spec 1.0

  • Support Nuclei RISC-V GCC 10.2


This is release 1.0.2 version of NMSIS-DSP library.

  • Sync up to CMSIS DSP library 1.9.0

  • Adding initial support for RISC-V vector extension support

  • Caution: riscv_math.h is separated into several header files. Extra PrivateInclude folder is included as header folder.


This is release V1.0.1 version of NMSIS-DSP library.

  • Both Nuclei RISC-V 32 and 64 bit cores are supported now.

  • Libraries are optimized for RISC-V 32 and 64 bit DSP instructions.

  • The NN examples are now using Nuclei SDK as running environment.


This is the first version of NMSIS-DSP library.

We adapt the CMSIS-DSP v1.6.0 library to use RISCV DSP instructions, all the API names now are renamed from arm_xxx to riscv_xxx.