CVTSD2SS

Convert Scalar Double Precision Floating-Point Value to Scalar Single Precision

stableVMJITAOTinstruction

Encodings

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.

Measured cost

Description#

Converts a double precision floating-point value in the "convert-from" source operand (the second operand in SSE2 version, otherwise the third operand) to a single precision floating-point value in the destination operand.

When the "convert-from" operand is an XMM register, the double precision floating-point value is contained in the low quadword of the register. The result is stored in the low doubleword of the destination operand. When the conversion is inexact, the value returned is rounded according to the rounding control bits in the MXCSR register.

128-bit Legacy SSE version: The "convert-from" source operand (the second operand) is an XMM register or memory location. Bits (MAXVL-1:32) of the corresponding destination register remain unchanged. The destination operand is an XMM register.

VEX.128 and EVEX encoded versions: The "convert-from" source operand (the third operand) can be an XMM register or a 64-bit memory location. The first source and destination operands are XMM registers. Bits (127:32) of the XMM register destination are copied from the corresponding bits in the first source operand. Bits (MAXVL- 1:128) of the destination register are zeroed.

EVEX encoded version: the converted result in written to the low doubleword element of the destination under the writemask.

Software should ensure VCVTSD2SS is encoded with VEX.L=0. Encoding VCVTSD2SS with VEX.L=1 may encounter unpredictable behavior across different processor generations.

Operation#

VCVTSD2SS (EVEX Encoded Version)

IF (SRC2 *is register*) AND (EVEX.b = 1)

     THEN

     SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(EVEX.RC);

     ELSE

     SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(MXCSR.RC);

FI;

IF k1[0] or *no writemask*

     THEN DEST[31:0] := Convert_Double_Precision_To_Single_Precision_Floating_Point(SRC2[63:0]);

     ELSE

     IF *merging-masking*                 ; merging-masking

           THEN *DEST[31:0] remains unchanged*

           ELSE                           ; zeroing-masking

           THEN DEST[31:0] := 0

     FI;

FI;

DEST[127:32] := SRC1[127:32]

DEST[MAXVL-1:128] := 0

VCVTSD2SS (VEX.128 Encoded Version)
DEST[31:0] := Convert_Double_Precision_To_Single_Precision_Floating_Point(SRC2[63:0]);
DEST[127:32] := SRC1[127:32]
DEST[MAXVL-1:128] := 0

CVTSD2SS (128-bit Legacy SSE Version)
DEST[31:0] := Convert_Double_Precision_To_Single_Precision_Floating_Point(SRC[63:0]);
(* DEST[MAXVL-1:32] Unmodified *)

Intel C/C++ compiler intrinsics#

VCVTSD2SS __m128 _mm_mask_cvtsd_ss(__m128 s, __mmask8 k, __m128 a, __m128d b);
VCVTSD2SS __m128 _mm_maskz_cvtsd_ss( __mmask8 k, __m128 a,__m128d b);
VCVTSD2SS __m128 _mm_cvt_roundsd_ss(__m128 a, __m128d b, int r);
VCVTSD2SS __m128 _mm_mask_cvt_roundsd_ss(__m128 s, __mmask8 k, __m128 a, __m128d b, int r);
VCVTSD2SS __m128 _mm_maskz_cvt_roundsd_ss( __mmask8 k, __m128 a,__m128d b, int r);
CVTSD2SS __m128_mm_cvtsd_ss(__m128 a, __m128d b);

SIMD Floating-Point Exceptions#

Overflow, Underflow, Invalid, Precision, Denormal.

Other Exceptions#

VEX-encoded instructions, see Table 2-20, "Type 3 Class Exception Conditions." EVEX-encoded instructions, see Table 2-49, "Type E3 Class Exception Conditions."

Sources

• This instruction in the Intel manual
• Latency and throughput per microarchitecture