Don't use -restrict for icpx

[masterleinad]

Summary

-restrict and -xHost are only supported by Intel® C++ Compiler Classic (https://www.intel.com/content/www/us/en/develop/documentation/cpp-compiler-developer-guide-and-reference/top/compiler-reference/compiler-options/compiler-option-details/language-options/restrict-qrestrict.html)

Related Issue(s)

Author(s)

Daniel Arndt, ORNL.

Licensing

By submitting this pull request, I agree, that my contribution will be included in LAMMPS and redistributed under either the GNU General Public License version 2 (GPL v2) or the GNU Lesser General Public License version 2.1 (LGPL v2.1).

Backward Compatibility

Yes.

Implementation Notes

Post Submission Checklist

• The feature or features in this pull request is complete
• Licensing information is complete
• Corresponding author information is complete
• The source code follows the LAMMPS formatting guidelines
• Suitable new documentation files and/or updates to the existing docs are included
• The added/updated documentation is integrated and tested with the documentation build system
• The feature has been verified to work with the conventional build system
• The feature has been verified to work with the CMake based build system
• Suitable tests have been added to the unittest tree.
• A package specific README file has been included or updated
• One or more example input decks are included

Further Information, Files, and Links

[junghans]

Not sure it is worth to make it conditional on the version.

[masterleinad]

Not sure it is worth to make it conditional on the version.

I'm not aware that anything changed for IntelLLVM compilers but I might be wrong.

[junghans]

Not sure it is worth to make it conditional on the version.

I'm not aware that anything changed for IntelLLVM compilers but I might be wrong.

If I recall correctly versions before 2021.4(?) support -restrict.

[junghans]

@masterleinad actually the bigger problem with the new icpx version is that floating point model is not strict anymore by default.
(see https://community.intel.com/t5/Intel-C-Compiler/icx-2021-3-0-bug-isinf-wrong-result/m-p/1316677/emcs_t/S2h8ZW1haWx8dG9waWNfc3Vic2NyaXB0aW9ufEtUV1BaMU43SzlPUEpQfDEzMTY2Nzd8U1VCU0NSSVBUSU9OU3xoSw#M39281)
which can lead to all kind of artifacts.

[masterleinad]

@masterleinad actually the bigger problem with the new icpx version is that floating point model is not strict anymore by default.
(see community.intel.com/t5/Intel-C-Compiler/icx-2021-3-0-bug-isinf-wrong-result/m-p/1316677/emcs_t/S2h8ZW1haWx8dG9waWNfc3Vic2NyaXB0aW9ufEtUV1BaMU43SzlPUEpQfDEzMTY2Nzd8U1VCU0NSSVBUSU9OU3xoSw#M39281)
which can lead to all kind of artifacts.

Yes, I know, but that's separate from some flags not being available at all.

[masterleinad]

@junghans Do you suggest forcing the floating-point model to strict or something the like?

[junghans]

@junghans Do you suggest forcing the floating-point model to strict or something the like?

Not sure, might be a huge performance hit, but of course nobody cares how fast you calculate the wrong answer. More a question for @akohlmey and @wmbrownIntel, but let's discuss this in another issue.