COMP: Forward CMAKE_APPLE_SILICON_PROCESSOR

[issakomi]

Forward CMAKE_APPLE_SILICON_PROCESSOR variable to external cmake configuration commands (Eigen3 and FFTW).
This fixes configuration for x86_64 on Apple Silicon, if CMAKE_APPLE_SILICON_PROCESSOR variable was set by user (value should be x86_64 or arm64).

Theoretically, for cross-compile setups, specially on other architectures, more variables should be forwarded, IMHO, but Eigen3 is header library, this is required only to pass correct architecture to "try compile" cmake's tests, AFAIK. And internal FFTW build is marked as non-production experimental build, so probably it is OK. Note: i didn't test FFTW build on Apple Silicon, but i think it is straightforward.

Let me know if you think CMAKE_OSX_ARCHITECTURES or other variables should be added too, so temporary set WIP.

S. comments to #2391

CC @seanm

Edit:
have tried "internal" FFTW builds (both variants, float and double) with CMAKE_APPLE_SILICON_PROCESSOR x86_64 on Apple Silicon, they are OK.

This variable exists in Cmake >= 3.19.2, BTW.

[thewtex]

Thanks, @issakomi !

[seanm]

Not sure I get the point of the existence of CMAKE_APPLE_SILICON_PROCESSOR, but it exists, so passing it down the line seems fine.

[issakomi]

Not sure I get the point of the existence of CMAKE_APPLE_SILICON_PROCESSOR, but it exists, so passing it down the line seems fine.

Here https://cmake.org/cmake/help/v3.19/release/3.19.html is some information:

Apple Silicon is now supported (since CMake 3.19.2):

3.19.2

• Apple Silicon host architecture selection support was updated. CMake 3.19.0 and 3.19.1 always chose arm64 as the host architecture. CMake 3.19.2 returns to using uname -m as CMake 3.18 and below did. Since this may vary based on CMake’s own architecture and that of the invoking process tree, the CMAKE_APPLE_SILICON_PROCESSOR variable or CMAKE_APPLE_SILICON_PROCESSOR environment variable may be set to specify a host architecture explicitly.

[issakomi]

@seanm

#include <cstdlib>
int main(int, char**)
{
	system("uname -a");
	return 0;
}

mi@mis-Mac-mini ~ % clang++ --target="x86_64-apple-macos10.14" test1.cpp
mi@mis-Mac-mini ~ % ./a.out
Darwin mis-Mac-mini.fritz.box 20.3.0 Darwin Kernel Version 20.3.0: Thu Jan 21 00:06:51 PST 2021; root:xnu-7195.81.3~1/RELEASE_ARM64_T8101 x86_64

mi@mis-Mac-mini ~ % clang++ test1.cpp
mi@mis-Mac-mini ~ % ./a.out
Darwin mis-Mac-mini.fritz.box 20.3.0 Darwin Kernel Version 20.3.0: Thu Jan 21 00:06:51 PST 2021; root:xnu-7195.81.3~1/RELEASE_ARM64_T8101 arm64

[seanm]

Ah, so uname gives different results running in rosetta or not. Makes sense I guess. It's after all trying to emulate an x86_64 and software that introspects uname results might expect to detect x86_64.