Struct U32x4

pub struct U32x4(/* private fields */);

[u32; 4] as a vector.

Implementations

impl U32x4

pub const fn from_array(arr: [u32; 4]) -> Self

Create a vector from an array.

Unlike the From trait function, the from_array function is const.

Example

const MY_EXTREMELY_FUN_VALUE: U32x4 = U32x4::from_array([0, 1, 2, 3]);
for (i, value) in MY_EXTREMELY_FUN_VALUE.as_array().iter().copied().enumerate() {
    assert_eq!(i as u32, value);
}

Trait Implementations

impl Add for U32x4

fn add(self, rhs: U32x4) -> U32x4

Perform a pairwise wrapping_add

Scalar Equivalent

U32x4::from([
    self.as_array()[0].wrapping_add(rhs.as_array()[0]),
    self.as_array()[1].wrapping_add(rhs.as_array()[1]),
    self.as_array()[2].wrapping_add(rhs.as_array()[2]),
    self.as_array()[3].wrapping_add(rhs.as_array()[3]),
])

AVX2 Intrinsics Used

• _mm_add_epi32
  • PADDD xmm, xmm

Neon Intrinsics Used

• vaddq_u32
  • This intrinsic compiles to the following instructions:
    • ADD Vd.4S,Vn.4S,Vm.4S

type Output = U32x4

The resulting type after applying the + operator.

impl AddAssign for U32x4

fn add_assign(&mut self, other: U32x4)

Performs the += operation.

impl AsMut<[u32]> for U32x4

fn as_mut(&mut self) -> &mut [u32]

Converts this type into a mutable reference of the (usually inferred) input type.

impl AsRef<[u32]> for U32x4

fn as_ref(&self) -> &[u32]

Converts this type into a shared reference of the (usually inferred) input type.

impl BitAnd for U32x4

fn bitand(self, rhs: U32x4) -> U32x4

Perform a pairwise bitwise and

Scalar Equivalent

U32x4::from([
    self.as_array()[0] & rhs.as_array()[0],
    self.as_array()[1] & rhs.as_array()[1],
    self.as_array()[2] & rhs.as_array()[2],
    self.as_array()[3] & rhs.as_array()[3],
])

AVX2 Intrinsics Used

• _mm_and_si128
  • PAND xmm, xmm

Neon Intrinsics Used

• vandq_u32
  • This intrinsic compiles to the following instructions:
    • AND Vd.16B,Vn.16B,Vm.16B

type Output = U32x4

The resulting type after applying the & operator.

impl BitAndAssign for U32x4

fn bitand_assign(&mut self, other: U32x4)

Performs the &= operation.

impl BitOr for U32x4

fn bitor(self, rhs: U32x4) -> U32x4

Perform a pairwise bitwise or

Scalar Equivalent

U32x4::from([
    self.as_array()[0] | rhs.as_array()[0],
    self.as_array()[1] | rhs.as_array()[1],
    self.as_array()[2] | rhs.as_array()[2],
    self.as_array()[3] | rhs.as_array()[3],
])

AVX2 Intrinsics Used

• _mm_or_si128
  • POR xmm, xmm

Neon Intrinsics Used

• vorrq_u32
  • This intrinsic compiles to the following instructions:
    • ORR Vd.16B,Vn.16B,Vm.16B

type Output = U32x4

The resulting type after applying the | operator.

impl BitOrAssign for U32x4

fn bitor_assign(&mut self, other: U32x4)

Performs the |= operation.

impl BitXor for U32x4

fn bitxor(self, rhs: U32x4) -> U32x4

Perform a pairwise bitwise xor

Scalar Equivalent

U32x4::from([
    self.as_array()[0] ^ rhs.as_array()[0],
    self.as_array()[1] ^ rhs.as_array()[1],
    self.as_array()[2] ^ rhs.as_array()[2],
    self.as_array()[3] ^ rhs.as_array()[3],
])

AVX2 Intrinsics Used

• _mm_xor_si128
  • PXOR xmm, xmm

Neon Intrinsics Used

• veorq_u32
  • This intrinsic compiles to the following instructions:
    • EOR Vd.16B,Vn.16B,Vm.16B

type Output = U32x4

The resulting type after applying the ^ operator.

impl BitXorAssign for U32x4

fn bitxor_assign(&mut self, other: U32x4)

Performs the ^= operation.

impl Clone for U32x4

fn clone(&self) -> U32x4

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl ConditionallySelectable for U32x4

fn conditional_select(a: &Self, b: &Self, choice: Choice) -> Self

Select a or b according to choice.

fn conditional_assign(&mut self, other: &Self, choice: Choice)

Conditionally assign other to self, according to choice.

fn conditional_swap(a: &mut Self, b: &mut Self, choice: Choice)

Conditionally swap self and other if choice == 1; otherwise, reassign both unto themselves.

impl ConstantTimeEq for U32x4

fn ct_eq(&self, other: &Self) -> Choice

Determine if two items are equal.

fn ct_ne(&self, other: &Self) -> Choice

Determine if two items are NOT equal.

impl Debug for U32x4

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for U32x4

fn default() -> Self

The zero vector.

impl<'de> Deserialize<'de> for U32x4

fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>

where D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Distribution<U32x4> for Standard

fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> U32x4

Generate a random value of T, using rng as the source of randomness.

fn sample_iter<R>(self, rng: R) -> DistIter<Self, R, T>

where R: Rng, Self: Sized,

Create an iterator that generates random values of T, using rng as the source of randomness.

fn map<F, S>(self, func: F) -> DistMap<Self, F, T, S>

where F: Fn(T) -> S, Self: Sized,

Create a distribution of values of ‘S’ by mapping the output of Self through the closure F

impl ExtendingCast<U16x8> for U32x4

fn extending_cast_from(vector: U16x8) -> U32x4

Scalar Equivalent:

U32x4::from([
        u32::from(vector.as_array()[0]),
        u32::from(vector.as_array()[1]),
        u32::from(vector.as_array()[2]),
        u32::from(vector.as_array()[3]),
])

Avx2

• _mm_cvtepu16_epi32

  • PMOVZXWD xmm, xmm

impl ExtendingCast<U32x4> for U64x2

fn extending_cast_from(vector: U32x4) -> U64x2

Scalar Equivalent:

U64x2::from([
        u64::from(vector.as_array()[0]),
        u64::from(vector.as_array()[1]),
])

Avx2

• _mm_cvtepu32_epi64

  • PMOVZXDQ xmm, xmm

impl ExtendingCast<U32x4> for U64x4

fn extending_cast_from(vector: U32x4) -> U64x4

Scalar Equivalent:

U64x4::from([
        u64::from(vector.as_array()[0]),
        u64::from(vector.as_array()[1]),
        u64::from(vector.as_array()[2]),
        u64::from(vector.as_array()[3]),
])

Avx2

• _mm256_cvtepu32_epi64

  • VPMOVZXDQ ymm, xmm

impl ExtendingCast<U8x16> for U32x4

fn extending_cast_from(vector: U8x16) -> U32x4

Scalar Equivalent:

U32x4::from([
        u32::from(vector.as_array()[0]),
        u32::from(vector.as_array()[1]),
        u32::from(vector.as_array()[2]),
        u32::from(vector.as_array()[3]),
])

Avx2

• _mm_cvtepu8_epi32

  • PMOVZXBD xmm, xmm

impl From<[u32; 4]> for U32x4

fn from(arr: [u32; 4]) -> U32x4

Converts to this type from the input type.

impl From<I16x8> for U32x4

fn from(x: I16x8) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of I16x8 as little endian bits of U32x4.

impl From<I32x4> for U32x4

fn from(x: I32x4) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of I32x4 as little endian bits of U32x4.

impl From<I64x2> for U32x4

fn from(x: I64x2) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of I64x2 as little endian bits of U32x4.

impl From<I8x16> for U32x4

fn from(x: I8x16) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of U32x4.

impl From<U16x8> for U32x4

fn from(x: U16x8) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of U16x8 as little endian bits of U32x4.

impl From<U32x4> for [u32; 4]

fn from(arr: U32x4) -> [u32; 4]

Converts to this type from the input type.

impl From<U32x4> for I16x8

fn from(x: U32x4) -> I16x8

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of I16x8.

impl From<U32x4> for I32x4

fn from(x: U32x4) -> I32x4

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of I32x4.

impl From<U32x4> for I64x2

fn from(x: U32x4) -> I64x2

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of I64x2.

impl From<U32x4> for I8x16

fn from(x: U32x4) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of I8x16.

impl From<U32x4> for U16x8

fn from(x: U32x4) -> U16x8

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of U16x8.

impl From<U32x4> for U32x8

fn from(vector: U32x4) -> U32x8

NOTE: this will zero the upper bits of the destination. Other intrinsics are more effcient, but leave the upper bits undefined. At present, these more effcient intrinsics are not exposed.

Scalar Equivalent:

let mut out = [0; 8];
out[0..4].copy_from_slice(&vector.as_array());
U32x8::from(out)

Avx2

• _mm256_zextsi128_si256

impl From<U32x4> for U64x2

fn from(x: U32x4) -> U64x2

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of U64x2.

impl From<U32x4> for U64x4

fn from(vector: U32x4) -> U64x4

Scalar Equivalent:

U64x4::from([
        u64::from(vector.as_array()[0]),
        u64::from(vector.as_array()[1]),
        u64::from(vector.as_array()[2]),
        u64::from(vector.as_array()[3]),
])

Avx2

• _mm256_cvtepu32_epi64

  • VPMOVZXDQ ymm, xmm

impl From<U32x4> for U8x16

fn from(x: U32x4) -> U8x16

This cast is 100% free. It reinterprets the little-endinan bits of U32x4 as little endian bits of U8x16.

impl From<U64x2> for U32x4

fn from(x: U64x2) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of U64x2 as little endian bits of U32x4.

impl From<U8x16> for U32x4

fn from(x: U8x16) -> U32x4

This cast is 100% free. It reinterprets the little-endinan bits of U8x16 as little endian bits of U32x4.

impl Hash for U32x4

fn hash<H: Hasher>(&self, state: &mut H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for U32x4

fn eq(&self, other: &Self) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for U32x4

fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>

where S: Serializer,

Serialize this value into the given Serde serializer.

impl Shl<u64> for U32x4

fn shl(self, amount: u64) -> U32x4

Scalar Equivalent:

if amount >= 32 {
    U32x4::ZERO
} else {
    U32x4::from([
        self.as_array()[0] << amount,
        self.as_array()[1] << amount,
        self.as_array()[2] << amount,
        self.as_array()[3] << amount,
    ])
}

Avx2

• _mm_set_epi64x

  Instruction sequence.

• _mm_sll_epi32

  • PSLLD xmm, xmm

type Output = U32x4

The resulting type after applying the << operator.

impl Shl for U32x4

fn shl(self, amount: U32x4) -> U32x4

Scalar Equivalent:

let mut out = self.as_array();
for (x, amm) in out.iter_mut().zip(amount.as_array().iter().copied()) {
    *x = if (0..32).contains(&amm) {
        *x << amm
    }  else {
        0
    };
}
U32x4::from(out)

Avx2

• _mm_sllv_epi32

  • VPSLLVD xmm, xmm, xmm

type Output = U32x4

The resulting type after applying the << operator.

impl ShlAssign<u64> for U32x4

fn shl_assign(&mut self, amount: u64)

Performs the <<= operation.

impl ShlAssign for U32x4

fn shl_assign(&mut self, amount: U32x4)

Performs the <<= operation.

impl Shr<u64> for U32x4

fn shr(self, amount: u64) -> U32x4

Scalar Equivalent:

if amount >= 32 {
    U32x4::ZERO
} else {
    U32x4::from([
        self.as_array()[0] >> amount,
        self.as_array()[1] >> amount,
        self.as_array()[2] >> amount,
        self.as_array()[3] >> amount,
    ])
}

Avx2

• _mm_set_epi64x

  Instruction sequence.

• _mm_srl_epi32

  • PSRLD xmm, xmm

type Output = U32x4

The resulting type after applying the >> operator.

impl Shr for U32x4

fn shr(self, amount: U32x4) -> U32x4

Scalar Equivalent:

let mut out = self.as_array();
for (x, amm) in out.iter_mut().zip(amount.as_array().iter().copied()) {
    *x = if (0..32).contains(&amm) {
        *x >> amm
    }  else {
        0
    };
}
U32x4::from(out)

Avx2

• _mm_srlv_epi32

  • VPSRLVD xmm, xmm, xmm

type Output = U32x4

The resulting type after applying the >> operator.

impl ShrAssign<u64> for U32x4

fn shr_assign(&mut self, amount: u64)

Performs the >>= operation.

impl ShrAssign for U32x4

fn shr_assign(&mut self, amount: U32x4)

Performs the >>= operation.

impl SimdBase for U32x4

fn is_zero(&self) -> bool

Scalar Equivalent:

self.as_array().iter().all(|x| *x == 0)

Avx2

• _mm_testz_si128

  • PTEST xmm, xmm

fn set_lo(scalar: u32) -> U32x4

Scalar Equivalent:

let mut out = [0; 4];
out[0] = scalar;
U32x4::from(out)

Avx2

• _mm_set_epi32

  Instruction sequence.

fn extract<const I: usize>(&self) -> u32

Scalar Equivalent:

self.as_array()[I]

Avx2

• _mm_extract_epi32

  • PEXTRD r32, xmm, imm8

fn broadcast(scalar: u32) -> U32x4

Scalar Equivalent:

U32x4::from([scalar; 4])

Avx2

• _mm_set1_epi32

  Instruction sequence.

fn broadcast_lo(vector: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([vector.as_array()[0]; 4])

Avx2

• _mm_broadcastd_epi32

  • VPBROADCASTD xmm, xmm

fn cmp_eq(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    if self.as_array()[0] == other.as_array()[0] {  u32::MAX  } else { 0 },
    if self.as_array()[1] == other.as_array()[1] {  u32::MAX  } else { 0 },
    if self.as_array()[2] == other.as_array()[2] {  u32::MAX  } else { 0 },
    if self.as_array()[3] == other.as_array()[3] {  u32::MAX  } else { 0 },
])

Avx2

• _mm_cmpeq_epi32

  • PCMPEQD xmm, xmm

fn and_not(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    self.as_array()[0] & (!other.as_array()[0]),
    self.as_array()[1] & (!other.as_array()[1]),
    self.as_array()[2] & (!other.as_array()[2]),
    self.as_array()[3] & (!other.as_array()[3]),
])

Avx2

• _mm_andnot_si128

  • PANDN xmm, xmm

fn cmp_gt(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    if self.as_array()[0] > other.as_array()[0] {  u32::MAX  } else { 0 },
    if self.as_array()[1] > other.as_array()[1] {  u32::MAX  } else { 0 },
    if self.as_array()[2] > other.as_array()[2] {  u32::MAX  } else { 0 },
    if self.as_array()[3] > other.as_array()[3] {  u32::MAX  } else { 0 },
])

Avx2

NOTE: this implementation uses an efficient vector polyfill, though this operation is not natively supported.

// Based on https://stackoverflow.com/a/33173643 and https://git.io/JmghK
let sign_bit = Self::broadcast(1 << 31);
Self::from(I32x4::from(*self ^ sign_bit).cmp_gt(
    I32x4::from(other ^ sign_bit)
))

fn shift_left<const BITS: usize>(&self) -> U32x4

Scalar Equivalent:

let mut out = self.as_array();
for x in out.iter_mut() {
    *x <<= BITS;
}
U32x4::from(out)

Avx2

• _mm_slli_epi32

  • PSLLD xmm, imm8

fn shift_right<const BITS: usize>(&self) -> U32x4

Scalar Equivalent:

let mut out = self.as_array();
for x in out.iter_mut() {
    *x >>= BITS;
}
U32x4::from(out)

Avx2

• _mm_srli_epi32

  • PSRLD xmm, imm8

fn unpack_lo(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    // Lane# 0
    self.as_array()[0],
    other.as_array()[0],
    self.as_array()[1],
    other.as_array()[1],
])

Avx2

• _mm_unpacklo_epi32

  • PUNPCKLDQ xmm, xmm

fn unpack_hi(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    // Lane# 0
    self.as_array()[2],
    other.as_array()[2],
    self.as_array()[3],
    other.as_array()[3],
])

Avx2

• _mm_unpackhi_epi32

  • PUNPCKHDQ xmm, xmm

fn max(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    self.as_array()[0].max(other.as_array()[0]),
    self.as_array()[1].max(other.as_array()[1]),
    self.as_array()[2].max(other.as_array()[2]),
    self.as_array()[3].max(other.as_array()[3]),
])

Avx2

• _mm_max_epu32

  • PMAXUD xmm, xmm

fn min(&self, other: U32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    self.as_array()[0].min(other.as_array()[0]),
    self.as_array()[1].min(other.as_array()[1]),
    self.as_array()[2].min(other.as_array()[2]),
    self.as_array()[3].min(other.as_array()[3]),
])

Avx2

• _mm_min_epu32

  • PMINUD xmm, xmm

const LANES: usize = 4usize

The number of elements of this vector.

const ZERO: Self

A vector where every element is zero.

type Scalar = u32

The scalar that this value holds.

type Array = [u32; 4]

The equivalent array type of this vector.

type Signed = I32x4

The signed version of this vector.

type Unsigned = U32x4

The unsigned version of this vector.

type BroadcastLoInput = U32x4

A vector of [Self::Scalar; 128 / (8 * std::mem::size_of::<Self::Scalar>())]

fn as_array(&self) -> Self::Array

Convert the vector to an array.

impl SimdBase32 for U32x4

fn shuffle<const I3: usize, const I2: usize, const I1: usize, const I0: usize>( &self, ) -> U32x4

Scalar Equivalent:

U32x4::from([
    // 128-bit Lane #0
    self.as_array()[I0 + 0 * 4],
    self.as_array()[I1 + 0 * 4],
    self.as_array()[I2 + 0 * 4],
    self.as_array()[I3 + 0 * 4],
])

Avx2

• _mm_shuffle_epi32

  • PSHUFD xmm, xmm, imm8

impl SimdBase4x for U32x4

fn blend<const B3: bool, const B2: bool, const B1: bool, const B0: bool>( &self, if_true: U32x4, ) -> U32x4

Scalar Equivalent:

U32x4::from([
        (if B0 { if_true } else { *self }).as_array()[0],
        (if B1 { if_true } else { *self }).as_array()[1],
        (if B2 { if_true } else { *self }).as_array()[2],
        (if B3 { if_true } else { *self }).as_array()[3],
])

Avx2

• _mm_blend_epi32

  • VPBLENDD xmm, xmm, xmm, imm8

impl SimdBaseGatherable<I32x4> for U32x4

Safety

base does not need to be aligned. Forall i, base + indices[i] must meet the safety requirements of std::ptr::read_unaligned

unsafe fn gather(base: *const u32, indices: I32x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    base.offset(indices.as_array()[0] as isize).read_unaligned(),
    base.offset(indices.as_array()[1] as isize).read_unaligned(),
    base.offset(indices.as_array()[2] as isize).read_unaligned(),
    base.offset(indices.as_array()[3] as isize).read_unaligned(),
])

Avx2

• _mm_i32gather_epi32

  • VPGATHERDD xmm, vm32x, xmm

unsafe fn gather_masked( base: *const u32, indices: I32x4, mask: U32x4, src: U32x4, ) -> U32x4

Scalar Equivalent:

U32x4::from([
    if (mask.as_array()[0] >> 31) == 1 {
        base.offset(indices.as_array()[0] as isize).read_unaligned()
    } else {
        src.as_array()[0]
    },
    if (mask.as_array()[1] >> 31) == 1 {
        base.offset(indices.as_array()[1] as isize).read_unaligned()
    } else {
        src.as_array()[1]
    },
    if (mask.as_array()[2] >> 31) == 1 {
        base.offset(indices.as_array()[2] as isize).read_unaligned()
    } else {
        src.as_array()[2]
    },
    if (mask.as_array()[3] >> 31) == 1 {
        base.offset(indices.as_array()[3] as isize).read_unaligned()
    } else {
        src.as_array()[3]
    },
])

Avx2

• _mm_mask_i32gather_epi32

  • VPGATHERDD xmm, vm32x, xmm

impl SimdBaseGatherable<I64x4> for U32x4

Safety

base does not need to be aligned. Forall i, base + indices[i] must meet the safety requirements of std::ptr::read_unaligned

unsafe fn gather(base: *const u32, indices: I64x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    base.offset(indices.as_array()[0] as isize).read_unaligned(),
    base.offset(indices.as_array()[1] as isize).read_unaligned(),
    base.offset(indices.as_array()[2] as isize).read_unaligned(),
    base.offset(indices.as_array()[3] as isize).read_unaligned(),
])

Avx2

• _mm256_i64gather_epi32

  • VPGATHERQD xmm, vm64y, xmm

unsafe fn gather_masked( base: *const u32, indices: I64x4, mask: U32x4, src: U32x4, ) -> U32x4

Scalar Equivalent:

U32x4::from([
    if (mask.as_array()[0] >> 31) == 1 {
        base.offset(indices.as_array()[0] as isize).read_unaligned()
    } else {
        src.as_array()[0]
    },
    if (mask.as_array()[1] >> 31) == 1 {
        base.offset(indices.as_array()[1] as isize).read_unaligned()
    } else {
        src.as_array()[1]
    },
    if (mask.as_array()[2] >> 31) == 1 {
        base.offset(indices.as_array()[2] as isize).read_unaligned()
    } else {
        src.as_array()[2]
    },
    if (mask.as_array()[3] >> 31) == 1 {
        base.offset(indices.as_array()[3] as isize).read_unaligned()
    } else {
        src.as_array()[3]
    },
])

Avx2

• _mm256_mask_i64gather_epi32

  • VPGATHERQD xmm, vm64y, xmm

impl SimdBaseGatherable<U64x4> for U32x4

Safety

base does not need to be aligned. Forall i, base + indices[i] must meet the safety requirements of std::ptr::read_unaligned

unsafe fn gather(base: *const u32, indices: U64x4) -> U32x4

Scalar Equivalent:

U32x4::from([
    base.offset(indices.as_array()[0] as isize).read_unaligned(),
    base.offset(indices.as_array()[1] as isize).read_unaligned(),
    base.offset(indices.as_array()[2] as isize).read_unaligned(),
    base.offset(indices.as_array()[3] as isize).read_unaligned(),
])

Avx2

• _mm256_i64gather_epi32

  • VPGATHERQD xmm, vm64y, xmm

unsafe fn gather_masked( base: *const u32, indices: U64x4, mask: U32x4, src: U32x4, ) -> U32x4

Scalar Equivalent:

U32x4::from([
    if (mask.as_array()[0] >> 31) == 1 {
        base.offset(indices.as_array()[0] as isize).read_unaligned()
    } else {
        src.as_array()[0]
    },
    if (mask.as_array()[1] >> 31) == 1 {
        base.offset(indices.as_array()[1] as isize).read_unaligned()
    } else {
        src.as_array()[1]
    },
    if (mask.as_array()[2] >> 31) == 1 {
        base.offset(indices.as_array()[2] as isize).read_unaligned()
    } else {
        src.as_array()[2]
    },
    if (mask.as_array()[3] >> 31) == 1 {
        base.offset(indices.as_array()[3] as isize).read_unaligned()
    } else {
        src.as_array()[3]
    },
])

Avx2

• _mm256_mask_i64gather_epi32

  • VPGATHERQD xmm, vm64y, xmm

impl Sub for U32x4

fn sub(self, rhs: U32x4) -> U32x4

Perform a pairwise wrapping_sub

Scalar Equivalent

U32x4::from([
    self.as_array()[0].wrapping_sub(rhs.as_array()[0]),
    self.as_array()[1].wrapping_sub(rhs.as_array()[1]),
    self.as_array()[2].wrapping_sub(rhs.as_array()[2]),
    self.as_array()[3].wrapping_sub(rhs.as_array()[3]),
])

AVX2 Intrinsics Used

• _mm_sub_epi32
  • PSUBD xmm, xmm

Neon Intrinsics Used

• vsubq_u32
  • This intrinsic compiles to the following instructions:
    • SUB Vd.4S,Vn.4S,Vm.4S

type Output = U32x4

The resulting type after applying the - operator.

impl SubAssign for U32x4

fn sub_assign(&mut self, other: U32x4)

Performs the -= operation.

impl Zeroable for U32x4

fn zeroed() -> Self

Calls zeroed.

impl Copy for U32x4

impl Eq for U32x4

impl Pod for U32x4