Would you like to use the trial-and-error approach so that you won't sacrifice the possibility of applying the optimization?

If there is a deterministic and cross-platform way to do this, I will be open to doing that.

GCC has a separate set of options for different architectures.
https://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html#Submodel-Options

It specifies -march=native in the arm options section, but it didn't work inside Godbolt.
https://gcc.gnu.org/onlinedocs/gcc/ARM-Options.html#ARM-Options

For Clang, I see the code for it here: Not sure why it is not enabled though
https://github.com/llvm/llvm-project/blob/17a68065c378da74805e4e1b9a5b78cc9f83e580/clang/lib/Driver/ToolChains/Arch/ARM.cpp#L893

For M1 Mac, it is under review:
https://reviews.llvm.org/D119788

If there is a deterministic and cross-platform way to do this, I will be open to doing that.

GCC has a separate set of options for different architectures. https://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html#Submodel-Options

It specifies -march=native in the arm options section, but it didn't work inside Godbolt. https://gcc.gnu.org/onlinedocs/gcc/ARM-Options.html#ARM-Options

I think a cross-platform solution without user input is almost impossible. A deterministic solution is also hard given the continuous (micro)architecture innovations for different CPUs, both GCC and clang are just keep incorporating the new changes. Therefore, I think a trial-and-error approach may be a better option for guiding in this situation, ie, test the flag before applying. I've tested on my ARM machine with the -mcpu=native on GCC and clang, both of them work.

As a relevant suggestion, we may set up an ARM-based environment using the Oracle A1 Instances (which says "Always Free") for better testing this project.

Yeah, if a trial and error solution is available I am open to adding it.

Changed #96 to reflect this.

Yeah, if a trial and error solution is available I am open to adding it.

For GCC and clang, we can use [gcc/clang] -Werror [the test flag] -E -xc /dev/null and check the return code.
For example, [gcc/clang] -Werror -march=native -E -xc /dev/null or [gcc/clang] -Werror -mcpu=native -E -xc /dev/null.

I guess MSVC will have a similar way to test, but I don't have the environment for testing.