How to use riscv+fpu？

[adfad1]

Hello!
I want to use fpu to accelerate computation, my c code is shown below, where "inputs" is defined as float.

I use this command to disassemble: make dis disopt="-d -l" | less, but i can not see any float instructions been used. What should I do to config the compiler?
Thanks in advance!

[bluewww]

You should try to compile with something like -march=rv32imfcxpulpv3

[adfad1]

You should try to compile with something like -march=rv32imfcxvpulpv3

I changed the Makefile like this

but error is reported

And I changed it into -march=rv32imfcxpulpv2 but there is still no float instructions

[bluewww]

I cannot reproduce. Please post a self contained c file and the compile command

[adfad1]

I cannot reproduce. Please post a self contained c file and the compile command

loss_function.tar.gz
I compile these codes under /pulpissimo/pulp-rt-examples/loss_function and didn't make any other changes on pulpissimo codes

[Ankasysemre]

use "PULP_CFLAGS = -O -g " in makefile.