GH-44249: [C++] Unify simd header includings

[zanmato1984]

Rationale for this change

#44249

What changes are included in this PR?

Change all the system header inclusions to simd.h.

Also removed some duplicated defines.

Are these changes tested?

Existing tests should be good.

Are there any user-facing changes?

None.

• GitHub Issue: [C++] Unifying simd header includings #44249

[github-actions]

⚠️ GitHub issue #44249 has been automatically assigned in GitHub to PR creator.

[zanmato1984]

Hi @pitrou , what do you think if we do this? And if you agree with this change, shall we move forward to add some lint rules to forbid system simd header including?

Thanks.

[zanmato1984]

@github-actions crossbow submit -g cpp

[github-actions]

Revision: d202127

Submitted crossbow builds: ursacomputing/crossbow @ actions-1cab231bec

Task	Status
example-cpp-minimal-build-static
example-cpp-minimal-build-static-system-dependency
example-cpp-tutorial
test-alpine-linux-cpp
test-build-cpp-fuzz
test-conda-cpp
test-conda-cpp-valgrind
test-cuda-cpp-ubuntu-20.04-cuda-11.2.2
test-cuda-cpp-ubuntu-22.04-cuda-11.7.1
test-debian-12-cpp-amd64
test-debian-12-cpp-i386
test-fedora-39-cpp
test-ubuntu-20.04-cpp
test-ubuntu-20.04-cpp-bundled
test-ubuntu-20.04-cpp-minimal-with-formats
test-ubuntu-22.04-cpp
test-ubuntu-22.04-cpp-20
test-ubuntu-22.04-cpp-emscripten
test-ubuntu-22.04-cpp-no-threading
test-ubuntu-24.04-cpp
test-ubuntu-24.04-cpp-gcc-13-bundled
test-ubuntu-24.04-cpp-gcc-14
test-ubuntu-24.04-cpp-thread-sanitizer

[zanmato1984]

Technically not related to the purpose of this change. Just saw it and fixed.

[zanmato1984]

And if you agree with this change, shall we move forward to add some lint rules to forbid system simd header including?

Thanks for looking at this @pitrou , much appreciated! And what do you think about this? The lint rule seems not trivial so I want to check with you first.

[pitrou]

And if you agree with this change, shall we move forward to add some lint rules to forbid system simd header including?

Thanks for looking at this @pitrou , much appreciated! And what do you think about this? The lint rule seems not trivial so I want to check with you first.

We could, yes, though I'm not sure how useful that would be. What would be the rationale?

[zanmato1984]

And if you agree with this change, shall we move forward to add some lint rules to forbid system simd header including?

Thanks for looking at this @pitrou , much appreciated! And what do you think about this? The lint rule seems not trivial so I want to check with you first.

We could, yes, though I'm not sure how useful that would be. What would be the rationale?

Right. After a second thought, I think the rational is not strong enough. One reason is that it doesn't really hurt anything if one uses system headers directly (except that one will have to write own platform-specific checking code) so maybe we shouldn't mandate that. The other is that intrinsics are not all simd related so requiring including simd.h (as opposed to including system intrinsic headers) for cases otherwise (like what we did for querying hardware flags) doesn't make much sense.

Let's just forget it for now. Thanks for reminding.

[pitrou]

The remaining CI failure is unrelated, thanks a lot @zanmato1984 !

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 60f29e9.

There was 1 benchmark result indicating a performance regression:

• Commit Run on test-mac-arm at 2024-10-02 10:22:41Z
  • tpch (R) with engine=arrow, format=native, language=R, memory_map=False, query_id=TPCH-18, scale_factor=1

The full Conbench report has more details. It also includes information about 6 possible false positives for unstable benchmarks that are known to sometimes produce them.