[Headers][X86] Allow vector bitcast intrinsics to be used in constexpr

[RKSimon]

Vector bitcast (and implicit truncation) cast intrinsics can all be used in constexpr.

_mm256_castpd_ps     _mm256_castps_pd
_mm512_castpd_ps     _mm512_castps_pd
_mm256_castpd_si256  _mm256_castps_si256
_mm512_castpd_si512  _mm512_castps_si512
_mm256_castsi256_pd  _mm512_castsi512_pd
_mm256_castsi256_ps  _mm512_castsi512_ps

_mm256_castpd256_pd128
_mm512_castpd512_pd128
_mm512_castpd512_pd256

_mm256_castps256_ps128
_mm512_castps512_ps128
_mm512_castps512_ps256
_mm256_castsi256_si128
_mm512_castsi512_si128
_mm512_castsi512_si256

_mm_castpd_ph _mm256_castpd_ph _mm512_castpd_ph
_mm_castph_pd _mm256_castph_pd _mm512_castph_pd
_mm_castph_ps _mm256_castph_ps _mm512_castph_ps
_mm_castps_ph _mm256_castps_ph _mm512_castps_ph

_mm_castph_si128 _mm256_castph_si256 _mm512_castph_si512
_mm_castsi128_ph _mm256_castsi256_ph _mm512_castsi512_ph

_mm256_castph256_ph128
_mm512_castph512_ph128
_mm512_castph512_ph256

Some of these might already have been done to support other constexpr intrinsics.

I may have missed some intrinsics in this category.....

NOTE the vector "any extension" casts (e.g. _mm256_castpd128_pd256) will have to stay non-constexpr for now as we can't use safely use __builtin_nondeterministic_value in constexpr.

[llvmbot]

Hi!

This issue may be a good introductory issue for people new to working on LLVM. If you would like to work on this issue, your first steps are:

1. Check that no other contributor is working on this issue. If someone is assigned to the issue or claimed to be working on it, ping the person. After one week without a response, the assignee may be changed.
2. Leave a comment indicating that you are working on the issue, or just create a pull request after following the steps below. Mention this issue in the description of the pull request.
3. Fix the issue locally.
4. Run the test suite locally. Remember that the subdirectories under test/ create fine-grained testing targets, so you can e.g. use make check-clang-ast to only run Clang's AST tests.
5. Create a Git commit.
6. Run git clang-format HEAD~1 to format your changes.
7. Open a pull request to the upstream repository on GitHub. Detailed instructions can be found in GitHub's documentation. Mention this issue in the description of the pull request.

If you have any further questions about this issue, don't hesitate to ask via a comment in the thread below.

[llvmbot]

@llvm/issue-subscribers-good-first-issue

Author: Simon Pilgrim (RKSimon)

Vector bitcast (and implicit truncation) cast intrinsics can all be used in constexpr. ``` _mm256_castpd_ps _mm256_castps_pd _mm512_castpd_ps _mm512_castps_pd _mm256_castpd_si256 _mm256_castps_si256 _mm512_castpd_si512 _mm512_castps_si512 _mm256_castsi256_pd _mm512_castsi512_pd _mm256_castsi256_ps _mm512_castsi512_ps

_mm256_castpd256_pd128
_mm512_castpd512_pd128
_mm512_castpd512_pd256

_mm256_castps256_ps128
_mm512_castps512_ps128
_mm512_castps512_ps256
_mm256_castsi256_si128
_mm512_castsi512_si128
_mm512_castsi512_si256

_mm_castpd_ph _mm256_castpd_ph _mm512_castpd_ph
_mm_castph_pd _mm256_castph_pd _mm512_castph_pd
_mm_castph_ps _mm256_castph_ps _mm512_castph_ps
_mm_castps_ph _mm256_castps_ph _mm512_castps_ph

_mm_castph_si128 _mm256_castph_si256 _mm512_castph_si512
_mm_castsi128_ph _mm256_castsi256_ph _mm512_castsi512_ph

_mm256_castph256_ph128
_mm512_castph512_ph128
_mm512_castph512_ph256

Some of these might already have been done to support other constexpr intrinsics.

I may have missed some intrinsics in this category.....

NOTE the vector "any extension" casts (e.g. _mm256_castpd128_pd256) will have to stay non-constexpr for now as we can't use safely use __builtin_nondeterministic_value in constexpr.
</details>

[llvmbot]

@llvm/issue-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

Vector bitcast (and implicit truncation) cast intrinsics can all be used in constexpr. ``` _mm256_castpd_ps _mm256_castps_pd _mm512_castpd_ps _mm512_castps_pd _mm256_castpd_si256 _mm256_castps_si256 _mm512_castpd_si512 _mm512_castps_si512 _mm256_castsi256_pd _mm512_castsi512_pd _mm256_castsi256_ps _mm512_castsi512_ps

_mm256_castpd256_pd128
_mm512_castpd512_pd128
_mm512_castpd512_pd256

_mm256_castps256_ps128
_mm512_castps512_ps128
_mm512_castps512_ps256
_mm256_castsi256_si128
_mm512_castsi512_si128
_mm512_castsi512_si256

_mm_castpd_ph _mm256_castpd_ph _mm512_castpd_ph
_mm_castph_pd _mm256_castph_pd _mm512_castph_pd
_mm_castph_ps _mm256_castph_ps _mm512_castph_ps
_mm_castps_ph _mm256_castps_ph _mm512_castps_ph

_mm_castph_si128 _mm256_castph_si256 _mm512_castph_si512
_mm_castsi128_ph _mm256_castsi256_ph _mm512_castsi512_ph

_mm256_castph256_ph128
_mm512_castph512_ph128
_mm512_castph512_ph256

Some of these might already have been done to support other constexpr intrinsics.

I may have missed some intrinsics in this category.....

NOTE the vector "any extension" casts (e.g. _mm256_castpd128_pd256) will have to stay non-constexpr for now as we can't use safely use __builtin_nondeterministic_value in constexpr.
</details>

[vedantjk]

Hi, can you take a look at my PR - #156381

[vedantjk]

@RKSimon I have a doubt, is there any methodology for making tests? Some casts tests might have really random values, is that ok?

[RKSimon]

Look at existing examples tests in the sse2-builtins.c

llvm-project/clang/test/CodeGen/X86/sse2-builtins.c

Lines 156 to 160 in 193df2a

	__m128 test_mm_castpd_ps(__m128d A) {
	// CHECK-LABEL: test_mm_castpd_ps
	return _mm_castpd_ps(A);
	}
	TEST_CONSTEXPR(match_m128(_mm_castpd_ps((__m128d){-1.0, +2.0}), +0.0f, -1.875f, +0.0f, +2.0f));

Its usually possible to create easily comparable fp values in one type from the raw bits of others - if all else fails we can match +/-0.0

The variants that cast between a si128/256/512 value <-> floating point value are much easier as at least one side can be represented by a raw hexvalue