AVX_ALL-Convert-XMM

AVX_ALL-Convert-XMM#

_mm_bcstnebf16_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const __bf16* __A

Param ETypes:

BF16 __A

__m128 _mm_bcstnebf16_ps(const __bf16* __A);

Intel Description

Convert scalar BF16 (16-bit) floating-point element stored at memory locations starting at location “__A” to a single-precision (32-bit) floating-point, broadcast it to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

b := Convert_BF16_To_FP32(MEM[__A+15:__A])
FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := b
ENDFOR
dst[MAX:128] := 0

_mm_bcstnesh_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const _Float16* __A

Param ETypes:

FP16 __A

__m128 _mm_bcstnesh_ps(const _Float16* __A);

Intel Description

Convert scalar half-precision (16-bit) floating-point element stored at memory locations starting at location “__A” to a single-precision (32-bit) floating-point, broadcast it to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

b := Convert_FP16_To_FP32(MEM[__A+15:__A])
FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := b
ENDFOR
dst[MAX:128] := 0

_mm_cvtneebf16_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const __m128bh* __A

Param ETypes:

BF16 __A

__m128 _mm_cvtneebf16_ps(const __m128bh* __A);

Intel Description

Convert packed BF16 (16-bit) floating-point even-indexed elements stored at memory locations starting at location “__A” to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := Convert_BF16_To_FP32(MEM[__A+m+15:__A+m])
ENDFOR
dst[MAX:128] := 0

_mm_cvtneeph_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const __m128h* __A

Param ETypes:

FP16 __A

__m128 _mm_cvtneeph_ps(const __m128h* __A);

Intel Description

Convert packed half-precision (16-bit) floating-point even-indexed elements stored at memory locations starting at location “__A” to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := Convert_FP16_To_FP32(MEM[__A+m+15:__A+m])
ENDFOR
dst[MAX:128] := 0

_mm_cvtneobf16_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const __m128bh* __A

Param ETypes:

BF16 __A

__m128 _mm_cvtneobf16_ps(const __m128bh* __A);

Intel Description

Convert packed BF16 (16-bit) floating-point odd-indexed elements stored at memory locations starting at location “__A” to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := Convert_BF16_To_FP32(MEM[__A+m+31:__A+m+16])
ENDFOR
dst[MAX:128] := 0

_mm_cvtneoph_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

const __m128h* __A

Param ETypes:

FP16 __A

__m128 _mm_cvtneoph_ps(const __m128h* __A);

Intel Description

Convert packed half-precision (16-bit) floating-point odd-indexed elements stored at memory locations starting at location “__A” to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        m := j*32
        dst[m+31:m] := Convert_FP16_To_FP32(MEM[__A+m+31:__A+m+16])
ENDFOR
dst[MAX:128] := 0

_mm_cvtneps_avx_pbh#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128bh

Param Types:

__m128 __A

Param ETypes:

FP32 __A

__m128bh _mm_cvtneps_avx_pbh(__m128 __A);

Intel Description

Convert packed single-precision (32-bit) floating-point elements in “__A” to packed BF16 (16-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        dst.word[j] := Convert_FP32_To_BF16(__A.fp32[j])
ENDFOR
dst[MAX:128] := 0

_mm_cvtneps_pbh#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128bh

Param Types:

__m128 __A

Param ETypes:

FP32 __A

__m128bh _mm_cvtneps_pbh(__m128 __A);

Intel Description

Convert packed single-precision (32-bit) floating-point elements in “__A” to packed BF16 (16-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        dst.word[j] := Convert_FP32_To_BF16(__A.fp32[j])
ENDFOR
dst[MAX:128] := 0

_mm_cvtph_ps#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128

Param Types:

__m128i a

Param ETypes:

FP16 a

__m128 _mm_cvtph_ps(__m128i a);

Intel Description

Convert packed half-precision (16-bit) floating-point elements in “a” to packed single-precision (32-bit) floating-point elements, and store the results in “dst”.

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        i := j*32
        m := j*16
        dst[i+31:i] := Convert_FP16_To_FP32(a[m+15:m])
ENDFOR
dst[MAX:128] := 0

_mm_cvtps_ph#

Tech:

AVX_ALL

Category:

Convert

Header:

immintrin.h

Searchable:

AVX_ALL-Convert-XMM

Register:

XMM 128 bit

Return Type:

__m128i

Param Types:

__m128 a, int imm8

Param ETypes:

FP32 a, IMM imm8

__m128i _mm_cvtps_ph(__m128 a, int imm8);

Intel Description

Convert packed single-precision (32-bit) floating-point elements in “a” to packed half-precision (16-bit) floating-point elements, and store the results in “dst”.

[round_imm_note]

Intel Implementation Psudeo-Code

FOR j := 0 to 3
        i := 16*j
        l := 32*j
        dst[i+15:i] := Convert_FP32_To_FP16(a[l+31:l])
ENDFOR
dst[MAX:64] := 0
Contents