VGETEXPPD—Convert Exponents of Packed DP FP Values to DP FP Values

Instruction Operand Encoding

Description

Extracts the biased exponents from the normalized DP FP representation of each qword data element of the source operand (the second operand) as unbiased signed integer value, or convert the denormal representation of input data to unbiased negative integer values. Each integer value of the unbiased exponent is converted to double-precision FP value and written to the corresponding qword elements of the destination operand (the first operand) as DP FP numbers.

The destination operand is a ZMM/YMM/XMM register and updated under the writemask. The source operand can be a ZMM/YMM/XMM register, a 512/256/128-bit memory location, or a 512/256/128-bit vector broadcasted from a 64-bit memory location.

EVEX.vvvv is reserved and must be 1111b, otherwise instructions will #UD.

Each GETEXP operation converts the exponent value into a FP number (permitting input value in denormal repre-sentation). Special cases of input values are listed in Table 5-7.

The formula is:

GETEXP(x) = floor(log₂(|x|))

Notation floor(x) stands for the greatest integer not exceeding real number x.

Table 5-7. VGETEXPPD/SD Special Cases

Operation

NormalizeExpTinyDPFP(SRC[63:0])

{

// Jbit is the hidden integral bit of a FP number. In case of denormal number it has the value of ZERO.

Src.Jbit (cid:197) 0;

Dst.exp (cid:197) 1;

Dst.fraction (cid:197) SRC[51:0];

WHILE(Src.Jbit = 0)

{

Src.Jbit (cid:197) Dst.fraction[51];

// Get the fraction MSB

Dst.fraction (cid:197) Dst.fraction << 1 ;

// One bit shift left

Dst.exp-- ;

// Decrement the exponent

}

Dst.fraction (cid:197) 0;

// zero out fraction bits

Dst.sign (cid:197) 1;

// Return negative sign

TMP[63:0] (cid:197) MXCSR.DAZ? 0 : (Dst.sign << 63) OR (Dst.exp << 52) OR (Dst.fraction) ;

Return (TMP[63:0]);

}

ConvertExpDPFP(SRC[63:0])

{

Src.sign (cid:197) 0;

// Zero out sign bit

Src.exp (cid:197) SRC[62:52];

Src.fraction (cid:197) SRC[51:0];

// Check for NaN

IF (SRC = NaN)

{

IF ( SRC = SNAN ) SET IE;

Return QNAN(SRC);

}

// Check for +INF

IF (SRC = +INF) Return (SRC);

// check if zero operand

IF ((Src.exp = 0) AND ((Src.fraction = 0) OR (MXCSR.DAZ = 1))) Return (-INF);

}

ELSE

// check if denormal operand (notice that MXCSR.DAZ = 0)

{

IF ((Src.exp = 0) AND (Src.fraction != 0))

{

TMP[63:0] (cid:197) NormalizeExpTinyDPFP(SRC[63:0]) ;

// Get Normalized Exponent

Set #DE

}

ELSE

// exponent value is correct

{

Dst.fraction (cid:197) 0;

// zero out fraction bits

TMP[63:0] (cid:197) (Src.sign << 63) OR (Src.exp << 52) OR (Src.fraction) ;

}

TMP (cid:197) SAR(TMP, 52) ;

// Shift Arithmetic Right

TMP (cid:197) TMP – 1023;

// Subtract Bias

Return CvtI2D(TMP) ;

// Convert INT to Double-Precision FP number

}

}

VGETEXPPD (EVEX encoded versions)

(KL, VL) = (2, 128), (4, 256), (8, 512)

FOR j (cid:197) 0 TO KL-1

i (cid:197) j * 64

IF k1[j] OR *no writemask*

THEN

IF (EVEX.b = 1) AND (SRC *is memory*)

THEN

DEST[i+63:i] (cid:197)

ConvertExpDPFP(SRC[63:0])

ELSE

DEST[i+63:i] (cid:197)

ConvertExpDPFP(SRC[i+63:i])

FI;

ELSE

IF *merging-masking*

; merging-masking

THEN *DEST[i+63:i] remains unchanged*

ELSE

; zeroing-masking

DEST[i+63:i] (cid:197) 0

FI

FI;

ENDFOR

DEST[MAX_VL-1:VL] (cid:197) 0

Intel C/C++ Compiler Intrinsic Equivalent

VGETEXPPD __m512d _mm512_getexp_pd(__m512d a);

VGETEXPPD __m512d _mm512_mask_getexp_pd(__m512d s, __mmask8 k, __m512d a);

VGETEXPPD __m512d _mm512_maskz_getexp_pd( __mmask8 k, __m512d a);

VGETEXPPD __m512d _mm512_getexp_round_pd(__m512d a, int sae);

VGETEXPPD __m512d _mm512_mask_getexp_round_pd(__m512d s, __mmask8 k, __m512d a, int sae);

VGETEXPPD __m512d _mm512_maskz_getexp_round_pd( __mmask8 k, __m512d a, int sae);

VGETEXPPD __m256d _mm256_getexp_pd(__m256d a);

VGETEXPPD __m256d _mm256_mask_getexp_pd(__m256d s, __mmask8 k, __m256d a);

VGETEXPPD __m256d _mm256_maskz_getexp_pd( __mmask8 k, __m256d a);

VGETEXPPD __m128d _mm_getexp_pd(__m128d a);

VGETEXPPD __m128d _mm_mask_getexp_pd(__m128d s, __mmask8 k, __m128d a);

VGETEXPPD __m128d _mm_maskz_getexp_pd( __mmask8 k, __m128d a);

SIMD Floating-Point Exceptions

Invalid, Denormal

Other Exceptions

See Exceptions Type E2.