CMake option and documentation for using the Intel compiler

[hakonsbm]

This PR adds a CMake option, with-intel, that is to be used when compiling with the Intel compiler, and documentation on this option. Adding

-Dwith-intel=ON

when configuring NEST adds the compiler flag -fp-model strict to ensure that computations obey the IEEE754 standard for floating point numerics. Other flags can also be set with

-Dwith-intel=<compiler-flags>

Fixes #242
Fixes #1027

[ikitayama]

Shouldn’t we do that in an arch/platform file instead the general ProcessOptions?

…

On Tue, Apr 9, 2019 at 22:31 Håkon Bakke Mørk ***@***.***> wrote: This PR adds a CMake option, with-intel, that is to be used when compiling with the Intel compiler, and documentation on this option. Adding -Dwith-intel=ON when configuring NEST adds the compiler flag -fp-model strict to ensure that computations obey the IEEE754 standard for floating point numerics. Other flags can also be set with -Dwith-intel=<compiler-flags> Fixes #242 <#242> Fixes #1027 <#1027> ------------------------------ You can view, comment on, or merge this pull request online at: #1170 Commit Summary - Added Cmake option for Intel compilers - Added documentation on Cmake option for the Intel compiler - Removed superfluous blank line File Changes - *M* CMakeLists.txt <https://github.com/nest/nest-simulator/pull/1170/files#diff-0> (2) - *M* INSTALL <https://github.com/nest/nest-simulator/pull/1170/files#diff-1> (3) - *M* cmake/ProcessOptions.cmake <https://github.com/nest/nest-simulator/pull/1170/files#diff-2> (12) - *M* doc/installation/install_options.rst <https://github.com/nest/nest-simulator/pull/1170/files#diff-3> (17) Patch Links: - https://github.com/nest/nest-simulator/pull/1170.patch - https://github.com/nest/nest-simulator/pull/1170.diff — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1170>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA6w5zSj5LKwu_hZlwGAiucyDE_AbMIhks5vfJZDgaJpZM4ckgku> .

[hakonsbm]

Some changes:

• -fp-model strict flag is set by default if the Intel compiler is used.
• This can be overridden with the option -Dwith-intel-flags=<compiler-flags>.

@ikitayama I'm not sure if we want a separate file just for compiler flags/options. But I'm happy to refactor if this is the decision we make.

[ikitayama]

It’s up to Hans’s call always.

…

On Thu, Apr 11, 2019 at 20:00 Håkon Bakke Mørk ***@***.***> wrote: Some changes: - -fp-model strict flag is set by default if the Intel compiler is used. - This can be overridden with the option -Dwith-intel-flags=<compiler-flags>. @ikitayama <https://github.com/ikitayama> I'm not sure if we want a separate file just for compiler flags/options. But I'm happy to refactor if this is the decision we make. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1170 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA6w54NjCzTfLMyw8c5vvkpFUG3Iw_lFks5vfxXYgaJpZM4ckgku> .

[jougs]

I'm happy with having this as a general option and would only introduce a platform file once we encounter a system that needs a bundling of several options.

For increased clarity, however, I would want to see --with-intel(-flags) renamed to --with-intel-compiler(-flags). As it is now, this could be understood as an option solely for Intel processors (and, e.g., excluding AMD or other CPUs).

[jougs]

Thanks for addressing my concerns and sorry for the long-ish delay for the re-review.

[gtrensch]

Thanks for this contribution! Looks good to me. 👍