MOVDDUP (Intel x86/64 assembly instruction)

모두의 코드
MOVDDUP (Intel x86/64 assembly instruction)

작성일 : 2020-09-01 이 글은 909 번 읽혔습니다.

MOVDDUP

Replicate Double FP Values

참고 사항

아래 표를 해석하는 방법은 x86-64 명령어 레퍼런스 읽는 법 글을 참조하시기 바랍니다.

Opcode/ Instruction	Op / En	64/32 bit Mode Support	CPUID Feature Flag	Description
`F2 0F 12 /r` MOVDDUP xmm1 xmm2/m64	RM	V/V	SSE3	Move double-precision floating-point value from xmm2/m64 and duplicate into xmm1.
`VEX.128.F2.0F.WIG 12 /r` VMOVDDUP xmm1 xmm2/m64	RM	V/V	AVX	Move double-precision floating-point value from xmm2/m64 and duplicate into xmm1.
`VEX.256.F2.0F.WIG 12 /r` VMOVDDUP ymm1 ymm2/m256	RM	V/V	AVX	Move even index double-precision floating-point values from ymm2/mem and duplicate each element into ymm1.
`EVEX.128.F2.0F.W1 12 /r` VMOVDDUP xmm1 {k1}{z} xmm2/m64	DUP-RM	V/V	AVX512VL AVX512F	Move double-precision floating-point value from xmm2/m64 and duplicate each element into xmm1 subject to writemask k1.
`EVEX.256.F2.0F.W1 12 /r` VMOVDDUP ymm1 {k1}{z} ymm2/m256	DUP-RM	V/V	AVX512VL AVX512F	Move even index double-precision floating-point values from ymm2/m256 and duplicate each element into ymm1 subject to writemask k1.
`EVEX.512.F2.0F.W1 12 /r` VMOVDDUP zmm1 {k1}{z} zmm2/m512	DUP-RM	V/V	AVX512F	Move even index double-precision floating-point values from zmm2/m512 and duplicate each element into zmm1 subject to writemask k1.

Instruction Operand Encoding

Op/En	Operand 1	Operand 2	Operand 3	Operand 4
RM	ModRM:reg (w)	ModRM:r/m (r)	NA	NA
DUP-RM	ModRM:reg (w)	ModRM:r/m (r)	NA	NA

Description

For 256-bit or higher versions: Duplicates even-indexed double-precision floating-point values from the source operand (the second operand) and into adjacent pair and store to the destination operand (the first operand).

For 128-bit versions: Duplicates the low double-precision floating-point value from the source operand (the second operand) and store to the destination operand (the first operand).

128-bit Legacy SSE version: Bits (MAXVL-1:128) of the corresponding destination register are unchanged. The source operand is XMM register or a 64-bit memory location.

VEX.128 and EVEX.128 encoded version: Bits (MAXVL-1:128) of the destination register are zeroed. The source operand is XMM register or a 64-bit memory location. The destination is updated conditionally under the writemask for EVEX version.

VEX.256 and EVEX.256 encoded version: Bits (MAXVL-1:256) of the destination register are zeroed. The source operand is YMM register or a 256-bit memory location. The destination is updated conditionally under the writemask for EVEX version.

EVEX.512 encoded version: The destination is updated according to the writemask. The source operand is ZMM register or a 512-bit memory location.

Note: VEX.vvvv and EVEX.vvvv are reserved and must be 1111b otherwise instructions will #UD.

Figure 4-2. VMOVDDUP Operation

Operation

VMOVDDUP (EVEX encoded versions)

(KL, VL) = (2, 128), (4, 256), (8, 512)
TMP_SRC[63:0] <-  SRC[63:0] 
TMP_SRC[127:64] <-  SRC[63:0]
IF VL >= 256
    TMP_SRC[191:128] <-  SRC[191:128]
    TMP_SRC[255:192] <-  SRC[191:128]
FI;
IF VL >= 512
    TMP_SRC[319:256] <-  SRC[319:256]
    TMP_SRC[383:320] <-  SRC[319:256]
    TMP_SRC[477:384] <-  SRC[477:384]
    TMP_SRC[511:484] <-  SRC[477:384]
FI;
FOR j <-  0 TO KL-1
    i <-  j * 64
    IF k1[j] OR *no writemask*
          THEN DEST[i+63:i] <-  TMP_SRC[i+63:i]
          ELSE 
                IF *merging-masking* ; merging-masking
                      THEN *DEST[i+63:i] remains unchanged*
                      ELSE  ; zeroing-masking
                            DEST[i+63:i] <-  0  ; zeroing-masking
                FI
    FI;
ENDFOR
DEST[MAX_VL-1:VL] <-  0

VMOVDDUP (VEX.256 encoded version)

DEST[63:0] <- SRC[63:0]
DEST[127:64] <- SRC[63:0]
DEST[191:128] <- SRC[191:128]
DEST[255:192] <- SRC[191:128]
DEST[MAX_VL-1:256] <- 0

VMOVDDUP (VEX.128 encoded version)

DEST[63:0] <- SRC[63:0]
DEST[127:64] <- SRC[63:0]
DEST[MAX_VL-1:128] <- 0

MOVDDUP (128-bit Legacy SSE version)

DEST[63:0] <- SRC[63:0]
DEST[127:64] <- SRC[63:0]
DEST[MAX_VL-1:128] (Unmodified)

Intel C/C++ Compiler Intrinsic Equivalent

VMOVDDUP __m512d _mm512_movedup_pd(__m512d a);
VMOVDDUP __m512d _mm512_mask_movedup_pd(__m512d s, __mmask8 k, __m512d a);
VMOVDDUP __m512d _mm512_maskz_movedup_pd(__mmask8 k, __m512d a);
VMOVDDUP __m256d _mm256_mask_movedup_pd(__m256d s, __mmask8 k, __m256d a);
VMOVDDUP __m256d _mm256_maskz_movedup_pd(__mmask8 k, __m256d a);
VMOVDDUP __m128d _mm_mask_movedup_pd(__m128d s, __mmask8 k, __m128d a);
VMOVDDUP __m128d _mm_maskz_movedup_pd(__mmask8 k, __m128d a);
MOVDDUP __m256d _mm256_movedup_pd(__m256d a);
MOVDDUP __m128d _mm_movedup_pd(__m128d a);

SIMD Floating-Point Exceptions

None

Other Exceptions

Non-EVEX-encoded instruction, see Exceptions Type 5;

EVEX-encoded instruction, see Exceptions Type E5NF.

#UD If EVEX.vvvv != 1111B or VEX.vvvv != 1111B.

첫 댓글을 달아주세요!

강좌에 관련 없이 궁금한 내용은 여기를 사용해주세요

또는 직접 입력하세요 (댓글 수정시 비밀번호가 필요합니다)

댓글을 불러오는 중입니다..

모두의 코드 MOVDDUP (Intel x86/64 assembly instruction)