This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Compiling CMSIS NN with GCC v12.2 for cortex-m55

Hi,

I'm trying to compile a simple project for cortex-m55, the project right now is pretty simple, just the main and CMSIS DSP and NN.

If I enable float-abi, no matter if hard, soft or softfp, I get the following compilation error:

../src/CMSIS/NN/Source/SoftmaxFunctions/arm_softmax_s8.c: In function 'arm_exp_on_negative_values_mve_32x4':

../src/CMSIS/NN/Source/SoftmaxFunctions/arm_softmax_s8.c:74:1: internal compiler error: in trunc_int_for_mode, at explow.cc:59

I have tried different combination like -mcpu=cortex-m55 or -march=armv8.1-m.main+fp.dp+mve.fp but the result is always the same.
What I'm doing wrong ? is there a way to compile for m55 with FPU enabled ? Otherwise MVE is not enabled.
Thx
fv
Parents
  • Thanks. This is a bug in GCC 12.2 that should be fixed in GCC 12.3 and GCC 13 when they are released. It looks like the latest CMSIS-NN sources have a workaround for this issue, so you could try updating your CMSIS-NN sources to see if that helps.

Reply
  • Thanks. This is a bug in GCC 12.2 that should be fixed in GCC 12.3 and GCC 13 when they are released. It looks like the latest CMSIS-NN sources have a workaround for this issue, so you could try updating your CMSIS-NN sources to see if that helps.

Children