Vesta
English
Source code

VGETEXPSD

Convert Exponents of Scalar Double Precision Floating-Point Value to Double

stableVMJITAOTinstruction

Encodings

OpcodeInstructionOp/En64-bitCompat/LegacyDescription
EVEX.LLIG.66.0F38.W1 43 /rVGETEXPSD xmm1 {k1}{z}, xmm2, xmm3/m64{sae}AValidValidConvert the biased exponent (bits 62:52) of the low OR AVX10.1 double precision floating-point value in xmm3/m64 to a double precision floating-point value representing unbiased integer exponent. Stores the result to the low 64-bit of xmm1 under the writemask k1 and merge with the other elements of xmm2.

Operand encoding

Each mode is a value of the Op/En column above. It says which field of the encoded instruction carries each operand, in the order they are written, and whether the instruction reads it, writes it or both.

A

  1. modrm.reg escrituraModRM byte, reg field (bits 5-3)
  2. evex.vvvv lecturaEVEX prefix, vvvv field (inverted)
  3. modrm.rm lecturaModRM byte, r/m field (bits 2-0); with the SIB byte and the displacement when the mod field asks for them

Tupla: Tuple1 Scalar

Measured cost

Loading measurements from arch-data...

Description

Extracts the biased exponent from the normalized double precision floating-point representation of the low qword data element of the source operand (the third operand) as unbiased signed integer value, or convert the denormal representation of input data to unbiased negative integer values. The integer value of the unbiased exponent is converted to double precision floating-point value and written to the destination operand (the first operand) as double precision floating-point numbers. Bits (127:64) of the XMM register destination are copied from corresponding bits in the first source operand.

The destination must be a XMM register, the source operand can be a XMM register or a float64 memory location.

If writemasking is used, the low quadword element of the destination operand is conditionally updated depending on the value of writemask register k1. If writemasking is not used, the low quadword element of the destination operand is unconditionally updated.

Each GETEXP operation converts the exponent value into a floating-point number (permitting input value in denormal representation). Special cases of input values are listed in Table 5-13.

The formula is:

GETEXP(x) = floor(log2(|x|)) Notation floor(x) stands for maximal integer not exceeding real number x.

Operation

// NormalizeExpTinyDPFP(SRC[63:0]) is defined in the Operation section of VGETEXPPD

// ConvertExpDPFP(SRC[63:0]) is defined in the Operation section of VGETEXPPD


VGETEXPSD (EVEX encoded version)

IF k1[0] OR *no writemask*

     THEN DEST[63:0] :=

           ConvertExpDPFP(SRC2[63:0])

     ELSE

     IF *merging-masking*              ; merging-masking

           THEN *DEST[63:0] remains unchanged*

           ELSE                        ; zeroing-masking

           DEST[63:0] := 0

     FI

FI;

DEST[127:64] := SRC1[127:64]

DEST[MAXVL-1:128] := 0

Intel C/C++ compiler intrinsics

VGETEXPSD __m128d _mm_getexp_sd( __m128d a, __m128d b);
VGETEXPSD __m128d _mm_mask_getexp_sd(__m128d s, __mmask8 k, __m128d a, __m128d b);
VGETEXPSD __m128d _mm_maskz_getexp_sd( __mmask8 k, __m128d a, __m128d b);
VGETEXPSD __m128d _mm_getexp_round_sd( __m128d a, __m128d b, int sae);
VGETEXPSD __m128d _mm_mask_getexp_round_sd(__m128d s, __mmask8 k, __m128d a, __m128d b, int sae);
VGETEXPSD __m128d _mm_maskz_getexp_round_sd( __mmask8 k, __m128d a, __m128d b, int sae);

SIMD Floating-Point Exceptions

Invalid, Denormal

Other Exceptions

See Table 2-49, "Type E3 Class Exception Conditions."

Sources