Use -march=haswell or similar flags instead of -march=native

[aminya]

The CMake file uses -march=native, which generates binaries that are not usable on other machines that might lack the instruction set.
I think the better approach would be to use -march=haswell, which is a good default these days.

For avx512, a CMake option could be exposed.

despacer/CMakeLists.txt

Lines 17 to 19 in 579530b

	target_compile_options(despacer PRIVATE /arch:native)
	else()
	target_compile_options(despacer PRIVATE -march=native)

-march=haswell includes AVX2 and older SSE instruction sets :
https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html

[sharpobject]

I think it makes a lot of sense to build this on the machine you intend to run it on.

[lemire]

@aminya

Intel recently disable AVX on some processors due to a vulnerability. So you cannot count on haswell to work on nearly all systems today.

The real solution is to use runtime dispatching.

[aminya]

There's also the libsimdpp, which already handles the runtime dispatch, and it could also fix the issue for supporting more architectures like Arm Neon (#4)

https://github.com/p12tic/libsimdpp

The dispatching part can be reused in case we don't want to rewrite the functions.
https://github.com/p12tic/libsimdpp/tree/master/simdpp/dispatch

[lemire]

Well, I know how to do the runtime dispatching. That's not a problem.

The question is whether anyone is planning to use this software for something serious. If so, they should get in touch with me and we can put in the effort.

I just wrote this code for research purposes.