[Headers][X86] Convert bf16 to f32 conversions to generic constexpr implementations

[RKSimon]

Now that we have the __bf16 type, the BF16 intrinsics can be converted to use standard conversions instead of bit twiddling that interferes with fp optimisations:
e.g.

static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(__bf16 __A) {
  return __builtin_ia32_cvtsbf162ss_32(__A); // remove __builtin_ia32_cvtsbf162ss_32
}

-->

static __inline__ float __DEFAULT_FN_ATTRS_CONSTEXPR _mm_cvtsbh_ss(__bf16 __A) {
  return (float)(__A);
}

and

static __inline__ __m512 __DEFAULT_FN_ATTRS512 _mm512_cvtpbh_ps(__m256bh __A) {
  return _mm512_castsi512_ps((__m512i)_mm512_slli_epi32(
      (__m512i)_mm512_cvtepi16_epi32((__m256i)__A), 16));
}

-->

static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR _mm512_cvtpbh_ps(__m256bh __A) {
  return (__m512)__builtin_convertvector(__A, __v16sf);
}

All the mask/maskz cvtpbh variants in avx512bf16intrin.h and avx512vlbf16intrin.h can be updated similarly

[llvmbot]

@llvm/issue-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

Now that we have the __bf16 type, the BF16 intrinsics can be converted to use standard conversions instead of bit twiddling that interferes with fp optimisations: e.g. ``` static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(__bf16 __A) { return __builtin_ia32_cvtsbf162ss_32(__A); // remove __builtin_ia32_cvtsbf162ss_32 } ``` --> ``` static __inline__ float __DEFAULT_FN_ATTRS_CONSTEXPR _mm_cvtsbh_ss(__bf16 __A) { return (float)(__A); } ``` and ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512 _mm512_cvtpbh_ps(__m256bh __A) { return _mm512_castsi512_ps((__m512i)_mm512_slli_epi32( (__m512i)_mm512_cvtepi16_epi32((__m256i)__A), 16)); } ``` --> ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR _mm512_cvtpbh_ps(__m256bh __A) { return (__m512)__builtin_convertvector(__A, __v16sf); } ``` All the mask/maskz cvtpbh variants in avx512bf16intrin.h and avx512vlbf16intrin.h can be updated similarly

[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)

Now that we have the __bf16 type, the BF16 intrinsics can be converted to use standard conversions instead of bit twiddling that interferes with fp optimisations: e.g. ``` static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(__bf16 __A) { return __builtin_ia32_cvtsbf162ss_32(__A); // remove __builtin_ia32_cvtsbf162ss_32 } ``` --> ``` static __inline__ float __DEFAULT_FN_ATTRS_CONSTEXPR _mm_cvtsbh_ss(__bf16 __A) { return (float)(__A); } ``` and ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512 _mm512_cvtpbh_ps(__m256bh __A) { return _mm512_castsi512_ps((__m512i)_mm512_slli_epi32( (__m512i)_mm512_cvtepi16_epi32((__m256i)__A), 16)); } ``` --> ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR _mm512_cvtpbh_ps(__m256bh __A) { return (__m512)__builtin_convertvector(__A, __v16sf); } ``` All the mask/maskz cvtpbh variants in avx512bf16intrin.h and avx512vlbf16intrin.h can be updated similarly

[llvmbot]

@llvm/issue-subscribers-clang-frontend

Author: Simon Pilgrim (RKSimon)

Now that we have the __bf16 type, the BF16 intrinsics can be converted to use standard conversions instead of bit twiddling that interferes with fp optimisations: e.g. ``` static __inline__ float __DEFAULT_FN_ATTRS _mm_cvtsbh_ss(__bf16 __A) { return __builtin_ia32_cvtsbf162ss_32(__A); // remove __builtin_ia32_cvtsbf162ss_32 } ``` --> ``` static __inline__ float __DEFAULT_FN_ATTRS_CONSTEXPR _mm_cvtsbh_ss(__bf16 __A) { return (float)(__A); } ``` and ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512 _mm512_cvtpbh_ps(__m256bh __A) { return _mm512_castsi512_ps((__m512i)_mm512_slli_epi32( (__m512i)_mm512_cvtepi16_epi32((__m256i)__A), 16)); } ``` --> ``` static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR _mm512_cvtpbh_ps(__m256bh __A) { return (__m512)__builtin_convertvector(__A, __v16sf); } ``` All the mask/maskz cvtpbh variants in avx512bf16intrin.h and avx512vlbf16intrin.h can be updated similarly

[wubowski]

Hello, I am new to the LLVM community and I would like to contribute! Can I be assigned to this issue?

[sivk-kex]

Hello, I would like to work on this issue.

[vso42]

Please assign it to me.

[RKSimon]

Sorry for the delay - its a public holiday in the UK - I've assigned it to @wubowski as they asked first

@sivk-kex and @vso42 would you be interested in splitting the work on #154539 ? one of you handles the int->half/float/double conversions, and the other the int<->int extensions/truncations?