vpsrlvd

Variable Bit Shift Right Logical Doubleword

VPSRLVD ymm1, ymm2, ymm3/m256

Shifts doublewords right logical by individual counts.

Details

Performs variable logical right shifts on packed 32-bit doublewords in the first source operand (ymm2) by the amounts specified in the corresponding doubleword elements of the second source operand (ymm3/m256), storing results in the destination (ymm1). Each doubleword is shifted independently; shift amounts > 31 produce zero. No flags are affected; this is an AVX2 instruction available in 256-bit form.

Pseudocode Operation

for i in 0 to 7:
  shift_amount = src2[i*32 : i*32+31]
  if shift_amount > 31:
    dest[i*32 : i*32+31] = 0
  else:
    dest[i*32 : i*32+31] = src1[i*32 : i*32+31] >> shift_amount

Example

VPSRLVD ymm1, ymm2, ymm3/m256

Encoding

Binary Layout
VEX
+0
opcode
+3
ModRM
+4
 
Format AVX2
Opcode VEX.256.66.0F38.W0 45 /r
Extension AVX2

Operands

  • dest
    256-bit YMM AVX register
  • src1
    256-bit YMM AVX register
  • src2
    256-bit YMM AVX register or Memory operand

Reference (Intel® SDM)

Instruction Forms

Opcode Instruction Op/En 64/32-bit Mode CPUID Description
VEX.128.66.0F38.W0 45 /r VPSRLVD xmm1, xmm2, xmm3/m128 A V/V AVX2 Shift doublewords in xmm2 right by amount specified in the corresponding element of xmm3/m128 while shifting in 0s.
VEX.128.66.0F38.W1 45 /r VPSRLVQ xmm1, xmm2, xmm3/m128 A V/V AVX2 Shift quadwords in xmm2 right by amount specified in the corresponding element of xmm3/m128 while shifting in 0s.
VEX.256.66.0F38.W0 45 /r VPSRLVD ymm1, ymm2, ymm3/m256 A V/V AVX2 Shift doublewords in ymm2 right by amount specified in the corresponding element of ymm3/m256 while shifting in 0s.
VEX.256.66.0F38.W1 45 /r VPSRLVQ ymm1, ymm2, ymm3/m256 A V/V AVX2 Shift quadwords in ymm2 right by amount specified in the corresponding element of ymm3/m256 while shifting in 0s.
EVEX.128.66.0F38.W1 10 /r VPSRLVW xmm1 {k1}{z}, xmm2, xmm3/m128 B V/V (AVX512VL AND AVX512BW) OR AVX10.1 Shift words in xmm2 right by amount specified in the corresponding element of xmm3/m128 while shifting in 0s using writemask k1.
EVEX.256.66.0F38.W1 10 /r VPSRLVW ymm1 {k1}{z}, ymm2, ymm3/m256 B V/V (AVX512VL AND AVX512BW) OR AVX10.1 Shift words in ymm2 right by amount specified in the corresponding element of ymm3/m256 while shifting in 0s using writemask k1.
EVEX.512.66.0F38.W1 10 /r VPSRLVW zmm1 {k1}{z}, zmm2, zmm3/m512 B V/V AVX512BW OR AVX10.1 Shift words in zmm2 right by amount specified in the corresponding element of zmm3/m512 while shifting in 0s using writemask k1.
EVEX.128.66.0F38.W0 45 /r VPSRLVD xmm1 {k1}{z}, xmm2, xmm3/m128/m32bcst C V/V (AVX512VL AND AVX512F) OR AVX10.1 Shift doublewords in xmm2 right by amount specified in the corresponding element of xmm3/m128/m32bcst while shifting in 0s using writemask k1.
EVEX.256.66.0F38.W0 45 /r VPSRLVD ymm1 {k1}{z}, ymm2, ymm3/m256/m32bcst C V/V (AVX512VL AND AVX512F) OR AVX10.1 Shift doublewords in ymm2 right by amount specified in the corresponding element of ymm3/m256/m32bcst while shifting in 0s using writemask k1.
EVEX.512.66.0F38.W0 45 /r VPSRLVD zmm1 {k1}{z}, zmm2, zmm3/m512/m32bcst C V/V AVX512F OR AVX10.1 Shift doublewords in zmm2 right by amount specified in the corresponding element of zmm3/m512/m32bcst while shifting in 0s using writemask k1.
EVEX.128.66.0F38.W1 45 /r VPSRLVQ xmm1 {k1}{z}, xmm2, xmm3/m128/m64bcst C V/V (AVX512VL AND AVX512F) OR AVX10.1 Shift quadwords in xmm2 right by amount specified in the corresponding element of xmm3/m128/m64bcst while shifting in 0s using writemask k1.
EVEX.256.66.0F38.W1 45 /r VPSRLVQ ymm1 {k1}{z}, ymm2, ymm3/m256/m64bcst C V/V (AVX512VL AND AVX512F) OR AVX10.1 Shift quadwords in ymm2 right by amount specified in the corresponding element of ymm3/m256/m64bcst while shifting in 0s using writemask k1.
EVEX.512.66.0F38.W1 45 /r VPSRLVQ zmm1 {k1}{z}, zmm2, zmm3/m512/m64bcst C V/V AVX512F OR AVX10.1 Shift quadwords in zmm2 right by amount specified in the corresponding element of zmm3/m512/m64bcst while shifting in 0s using writemask k1. VPSRLVW/VPSRLVD/VPSRLVQ—Variable Bit Shift Right Logical Vol. 2C 5-646

Instruction Operand Encoding

Op/En Tuple Type Operand 1 Operand 2 Operand 3 Operand 4
A N/A ModRM:reg (w) VEX.vvvv (r) ModRM:r/m (r) N/A
B Full Mem ModRM:reg (w) EVEX.vvvv (r) ModRM:r/m (r) N/A
C Full ModRM:reg (w) EVEX.vvvv (r) ModRM:r/m (r) N/A

Description

Shifts the bits in the individual data elements (words, doublewords or quadword) in the first source operand to the right by the count value of respective data elements in the second source operand. As the bits in the data elements are shifted right, the empty high-order bits are cleared (set to 0). The count values are specified individually in each data element of the second source operand. If the unsigned integer value specified in the respective data element of the second source operand is greater than 15 (for word), 31 (for doublewords), or 63 (for a quadword), then the destination data element are written with 0. VEX.128 encoded version: The destination and first source operands are XMM registers. The count operand can be either an XMM register or a 128-bit memory location. Bits (MAXVL-1:128) of the corresponding destination register are zeroed. VEX.256 encoded version: The destination and first source operands are YMM registers. The count operand can be either an YMM register or a 256-bit memory. Bits (MAXVL-1:256) of the corresponding ZMM register are zeroed. EVEX encoded VPSRLVD/Q: The destination and first source operands are ZMM/YMM/XMM registers. The count operand can be either a ZMM/YMM/XMM register, a 512/256/128-bit memory location or a 512-bit vector broadcasted from a 32/64-bit memory location. The destination is conditionally updated with writemask k1. EVEX encoded VPSRLVW: The destination and first source operands are ZMM/YMM/XMM registers. The count operand can be either a ZMM/YMM/XMM register, a 512/256/128-bit memory location. The destination is conditionally updated with writemask k1.

Operation

VPSRLVW (EVEX encoded version)
(KL, VL) = (8, 128), (16, 256), (32, 512)

FOR j := 0 TO KL-1
i := j * 16
IF k1[j] OR *no writemask*
THEN DEST[i+15:i] := ZeroExtend(SRC1[i+15:i] >> SRC2[i+15:i])
ELSE
IF *merging-masking*                                 ; merging-masking
THEN *DEST[i+15:i] remains unchanged*
ELSE                                                         ; zeroing-masking
DEST[i+15:i] := 0
FI
FI;
ENDFOR;
DEST[MAXVL-1:VL] := 0;





VPSRLVW/VPSRLVD/VPSRLVQ—Variable Bit Shift Right Logical                                                                                Vol. 2C 5-647
VPSRLVD (VEX.128 version)
COUNT_0 := SRC2[31 : 0]
(* Repeat Each COUNT_i for the 2nd through 4th dwords of SRC2*)
COUNT_3 := SRC2[127 : 96];
IF COUNT_0 < 32 THEN
DEST[31:0] := ZeroExtend(SRC1[31:0] >> COUNT_0);
ELSE
DEST[31:0] := 0;
(* Repeat shift operation for 2nd through 4th dwords *)
IF COUNT_3 < 32 THEN
DEST[127:96] := ZeroExtend(SRC1[127:96] >> COUNT_3);
ELSE
DEST[127:96] := 0;
DEST[MAXVL-1:128] := 0;

VPSRLVD (VEX.256 version)
COUNT_0 := SRC2[31 : 0];
(* Repeat Each COUNT_i for the 2nd through 7th dwords of SRC2*)
COUNT_7 := SRC2[255 : 224];
IF COUNT_0 < 32 THEN
DEST[31:0] := ZeroExtend(SRC1[31:0] >> COUNT_0);
ELSE
DEST[31:0] := 0;
(* Repeat shift operation for 2nd through 7th dwords *)
IF COUNT_7 < 32 THEN
DEST[255:224] := ZeroExtend(SRC1[255:224] >> COUNT_7);
ELSE
DEST[255:224] := 0;
DEST[MAXVL-1:256] := 0;

VPSRLVD (EVEX encoded version)
(KL, VL) = (4, 128), (8, 256), (16, 512)
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask* THEN
IF (EVEX.b = 1) AND (SRC2 *is memory*)
THEN DEST[i+31:i] := ZeroExtend(SRC1[i+31:i] >> SRC2[31:0])
ELSE DEST[i+31:i] := ZeroExtend(SRC1[i+31:i] >> SRC2[i+31:i])
FI;
ELSE
IF *merging-masking*                                 ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE                                                         ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR;
DEST[MAXVL-1:VL] := 0;





VPSRLVW/VPSRLVD/VPSRLVQ—Variable Bit Shift Right Logical                                                                                Vol. 2C 5-648
VPSRLVQ (VEX.128 version)
COUNT_0 := SRC2[63 : 0];
COUNT_1 := SRC2[127 : 64];
IF COUNT_0 < 64 THEN
DEST[63:0] := ZeroExtend(SRC1[63:0] >> COUNT_0);
ELSE
DEST[63:0] := 0;
IF COUNT_1 < 64 THEN
DEST[127:64] := ZeroExtend(SRC1[127:64] >> COUNT_1);
ELSE
DEST[127:64] := 0;
DEST[MAXVL-1:128] := 0;

VPSRLVQ (VEX.256 version)
COUNT_0 := SRC2[63 : 0];
(* Repeat Each COUNT_i for the 2nd through 4th dwords of SRC2*)
COUNT_3 := SRC2[255 : 192];
IF COUNT_0 < 64 THEN
DEST[63:0] := ZeroExtend(SRC1[63:0] >> COUNT_0);
ELSE
DEST[63:0] := 0;
(* Repeat shift operation for 2nd through 4th dwords *)
IF COUNT_3 < 64 THEN
DEST[255:192] := ZeroExtend(SRC1[255:192] >> COUNT_3);
ELSE
DEST[255:192] := 0;
DEST[MAXVL-1:256] := 0;

VPSRLVQ (EVEX encoded version)
(KL, VL) = (2, 128), (4, 256), (8, 512)
FOR j := 0 TO KL-1
i := j * 64
IF k1[j] OR *no writemask* THEN
IF (EVEX.b = 1) AND (SRC2 *is memory*)
THEN DEST[i+63:i] := ZeroExtend(SRC1[i+63:i] >> SRC2[63:0])
ELSE DEST[i+63:i] := ZeroExtend(SRC1[i+63:i] >> SRC2[i+63:i])
FI;
ELSE
IF *merging-masking*                                 ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE                                                         ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR;
DEST[MAXVL-1:VL] := 0;





VPSRLVW/VPSRLVD/VPSRLVQ—Variable Bit Shift Right Logical                                                                                Vol. 2C 5-649

Intel C/C++ Compiler Intrinsic Equivalent

VPSRLVW __m512i _mm512_srlv_epi16(__m512i a, __m512i cnt);
VPSRLVW __m512i _mm512_mask_srlv_epi16(__m512i s, __mmask32 k, __m512i a, __m512i cnt);
VPSRLVW __m512i _mm512_maskz_srlv_epi16( __mmask32 k, __m512i a, __m512i cnt);
VPSRLVW __m256i _mm256_mask_srlv_epi16(__m256i s, __mmask16 k, __m256i a, __m256i cnt);
VPSRLVW __m256i _mm256_maskz_srlv_epi16( __mmask16 k, __m256i a, __m256i cnt);
VPSRLVW __m128i _mm_mask_srlv_epi16(__m128i s, __mmask8 k, __m128i a, __m128i cnt);
VPSRLVW __m128i _mm_maskz_srlv_epi16( __mmask8 k, __m128i a, __m128i cnt);
VPSRLVW __m256i _mm256_srlv_epi32 (__m256i m, __m256i count)
VPSRLVD __m512i _mm512_srlv_epi32(__m512i a, __m512i cnt);
VPSRLVD __m512i _mm512_mask_srlv_epi32(__m512i s, __mmask16 k, __m512i a, __m512i cnt);
VPSRLVD __m512i _mm512_maskz_srlv_epi32( __mmask16 k, __m512i a, __m512i cnt);
VPSRLVD __m256i _mm256_mask_srlv_epi32(__m256i s, __mmask8 k, __m256i a, __m256i cnt);
VPSRLVD __m256i _mm256_maskz_srlv_epi32( __mmask8 k, __m256i a, __m256i cnt);
VPSRLVD __m128i _mm_mask_srlv_epi32(__m128i s, __mmask8 k, __m128i a, __m128i cnt);
VPSRLVD __m128i _mm_maskz_srlv_epi32( __mmask8 k, __m128i a, __m128i cnt);
VPSRLVQ __m512i _mm512_srlv_epi64(__m512i a, __m512i cnt);
VPSRLVQ __m512i _mm512_mask_srlv_epi64(__m512i s, __mmask8 k, __m512i a, __m512i cnt);
VPSRLVQ __m512i _mm512_maskz_srlv_epi64( __mmask8 k, __m512i a, __m512i cnt);
VPSRLVQ __m256i _mm256_mask_srlv_epi64(__m256i s, __mmask8 k, __m256i a, __m256i cnt);
VPSRLVQ __m256i _mm256_maskz_srlv_epi64( __mmask8 k, __m256i a, __m256i cnt);
VPSRLVQ __m128i _mm_mask_srlv_epi64(__m128i s, __mmask8 k, __m128i a, __m128i cnt);
VPSRLVQ __m128i _mm_maskz_srlv_epi64( __mmask8 k, __m128i a, __m128i cnt);
VPSRLVQ __m256i _mm256_srlv_epi64 (__m256i m, __m256i count)
VPSRLVD __m128i _mm_srlv_epi32( __m128i a, __m128i cnt);
VPSRLVQ __m128i _mm_srlv_epi64( __m128i a, __m128i cnt);

Exceptions

SIMD Floating-Point Exceptions

None.

Other Exceptions

VEX-encoded instructions, see Table 2-21, “Type 4 Class Exception Conditions.” EVEX-encoded VPSRLVD/Q, see Table 2-51, “Type E4 Class Exception Conditions.” EVEX-encoded VPSRLVW, see Exceptions Type E4.nb in Table 2-51, “Type E4 Class Exception Conditions.” VPSRLVW/VPSRLVD/VPSRLVQ—Variable Bit Shift Right Logical Vol. 2C 5-650