Struct I8x16

pub struct I8x16(/* private fields */);

[i8; 16] as a vector.

Implementations

impl I8x16

pub const fn from_array(arr: [i8; 16]) -> Self

Create a vector from an array.

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

Example

const MY_EXTREMELY_FUN_VALUE: I8x16 = I8x16::from_array([
    0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
]);
for (i, value) in MY_EXTREMELY_FUN_VALUE.as_array().iter().copied().enumerate() {
    assert_eq!(i as i8, value);
}

impl I8x16

pub fn shuffle(&self, order: U8x16) -> I8x16

Scalar Equivalent:

let mut arr = [0; 16];
for (lane_dst, (lane_src, order)) in
    arr.chunks_exact_mut(16).zip(
        self.as_array().chunks_exact(16)
        .zip(order.as_array().chunks_exact(16))
    )
{
    for (dst, idx) in lane_dst.iter_mut().zip(order) {
        let idx = *idx;
        *dst = if (idx >> 7) == 1 {
            0
        } else {
            lane_src[(idx as usize) % 16]
        };
    }
}
arr.into()

Avx2

• _mm_shuffle_epi8

  • PSHUFB xmm, xmm

Trait Implementations

impl Add for I8x16

fn add(self, rhs: I8x16) -> I8x16

Perform a pairwise wrapping_add

Scalar Equivalent

I8x16::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]),
    self.as_array()[4].wrapping_add(rhs.as_array()[4]),
    self.as_array()[5].wrapping_add(rhs.as_array()[5]),
    self.as_array()[6].wrapping_add(rhs.as_array()[6]),
    self.as_array()[7].wrapping_add(rhs.as_array()[7]),
    self.as_array()[8].wrapping_add(rhs.as_array()[8]),
    self.as_array()[9].wrapping_add(rhs.as_array()[9]),
    self.as_array()[10].wrapping_add(rhs.as_array()[10]),
    self.as_array()[11].wrapping_add(rhs.as_array()[11]),
    self.as_array()[12].wrapping_add(rhs.as_array()[12]),
    self.as_array()[13].wrapping_add(rhs.as_array()[13]),
    self.as_array()[14].wrapping_add(rhs.as_array()[14]),
    self.as_array()[15].wrapping_add(rhs.as_array()[15]),
])

AVX2 Intrinsics Used

• _mm_add_epi8
  • PADDB xmm, xmm

Neon Intrinsics Used

• vaddq_s8
  • This intrinsic compiles to the following instructions:
    • ADD Vd.16B,Vn.16B,Vm.16B

type Output = I8x16

The resulting type after applying the + operator.

impl AddAssign for I8x16

fn add_assign(&mut self, other: I8x16)

Performs the += operation.

impl AsMut<[i8]> for I8x16

fn as_mut(&mut self) -> &mut [i8]

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

impl AsRef<[i8]> for I8x16

fn as_ref(&self) -> &[i8]

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

impl BitAnd for I8x16

fn bitand(self, rhs: I8x16) -> I8x16

Perform a pairwise bitwise and

Scalar Equivalent

I8x16::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],
    self.as_array()[4] & rhs.as_array()[4],
    self.as_array()[5] & rhs.as_array()[5],
    self.as_array()[6] & rhs.as_array()[6],
    self.as_array()[7] & rhs.as_array()[7],
    self.as_array()[8] & rhs.as_array()[8],
    self.as_array()[9] & rhs.as_array()[9],
    self.as_array()[10] & rhs.as_array()[10],
    self.as_array()[11] & rhs.as_array()[11],
    self.as_array()[12] & rhs.as_array()[12],
    self.as_array()[13] & rhs.as_array()[13],
    self.as_array()[14] & rhs.as_array()[14],
    self.as_array()[15] & rhs.as_array()[15],
])

AVX2 Intrinsics Used

• _mm_and_si128
  • PAND xmm, xmm

Neon Intrinsics Used

• vandq_s8
  • This intrinsic compiles to the following instructions:
    • AND Vd.16B,Vn.16B,Vm.16B

type Output = I8x16

The resulting type after applying the & operator.

impl BitAndAssign for I8x16

fn bitand_assign(&mut self, other: I8x16)

Performs the &= operation.

impl BitOr for I8x16

fn bitor(self, rhs: I8x16) -> I8x16

Perform a pairwise bitwise or

Scalar Equivalent

I8x16::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],
    self.as_array()[4] | rhs.as_array()[4],
    self.as_array()[5] | rhs.as_array()[5],
    self.as_array()[6] | rhs.as_array()[6],
    self.as_array()[7] | rhs.as_array()[7],
    self.as_array()[8] | rhs.as_array()[8],
    self.as_array()[9] | rhs.as_array()[9],
    self.as_array()[10] | rhs.as_array()[10],
    self.as_array()[11] | rhs.as_array()[11],
    self.as_array()[12] | rhs.as_array()[12],
    self.as_array()[13] | rhs.as_array()[13],
    self.as_array()[14] | rhs.as_array()[14],
    self.as_array()[15] | rhs.as_array()[15],
])

AVX2 Intrinsics Used

• _mm_or_si128
  • POR xmm, xmm

Neon Intrinsics Used

• vorrq_s8
  • This intrinsic compiles to the following instructions:
    • ORR Vd.16B,Vn.16B,Vm.16B

type Output = I8x16

The resulting type after applying the | operator.

impl BitOrAssign for I8x16

fn bitor_assign(&mut self, other: I8x16)

Performs the |= operation.

impl BitXor for I8x16

fn bitxor(self, rhs: I8x16) -> I8x16

Perform a pairwise bitwise xor

Scalar Equivalent

I8x16::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],
    self.as_array()[4] ^ rhs.as_array()[4],
    self.as_array()[5] ^ rhs.as_array()[5],
    self.as_array()[6] ^ rhs.as_array()[6],
    self.as_array()[7] ^ rhs.as_array()[7],
    self.as_array()[8] ^ rhs.as_array()[8],
    self.as_array()[9] ^ rhs.as_array()[9],
    self.as_array()[10] ^ rhs.as_array()[10],
    self.as_array()[11] ^ rhs.as_array()[11],
    self.as_array()[12] ^ rhs.as_array()[12],
    self.as_array()[13] ^ rhs.as_array()[13],
    self.as_array()[14] ^ rhs.as_array()[14],
    self.as_array()[15] ^ rhs.as_array()[15],
])

AVX2 Intrinsics Used

• _mm_xor_si128
  • PXOR xmm, xmm

Neon Intrinsics Used

• veorq_s8
  • This intrinsic compiles to the following instructions:
    • EOR Vd.16B,Vn.16B,Vm.16B

type Output = I8x16

The resulting type after applying the ^ operator.

impl BitXorAssign for I8x16

fn bitxor_assign(&mut self, other: I8x16)

Performs the ^= operation.

impl Clone for I8x16

fn clone(&self) -> I8x16

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl ConditionallySelectable for I8x16

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 I8x16

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 I8x16

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

Formats the value using the given formatter.

impl Default for I8x16

fn default() -> Self

The zero vector.

impl<'de> Deserialize<'de> for I8x16

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

where D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Distribution<I8x16> for Standard

fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> I8x16

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<I8x16> for I16x16

fn extending_cast_from(vector: I8x16) -> I16x16

Scalar Equivalent:

I16x16::from([
        i16::from(vector.as_array()[0]),
        i16::from(vector.as_array()[1]),
        i16::from(vector.as_array()[2]),
        i16::from(vector.as_array()[3]),
        i16::from(vector.as_array()[4]),
        i16::from(vector.as_array()[5]),
        i16::from(vector.as_array()[6]),
        i16::from(vector.as_array()[7]),
        i16::from(vector.as_array()[8]),
        i16::from(vector.as_array()[9]),
        i16::from(vector.as_array()[10]),
        i16::from(vector.as_array()[11]),
        i16::from(vector.as_array()[12]),
        i16::from(vector.as_array()[13]),
        i16::from(vector.as_array()[14]),
        i16::from(vector.as_array()[15]),
])

Avx2

• _mm256_cvtepi8_epi16

  • VPMOVSXBW ymm, xmm

impl ExtendingCast<I8x16> for I16x8

fn extending_cast_from(vector: I8x16) -> I16x8

Scalar Equivalent:

I16x8::from([
        i16::from(vector.as_array()[0]),
        i16::from(vector.as_array()[1]),
        i16::from(vector.as_array()[2]),
        i16::from(vector.as_array()[3]),
        i16::from(vector.as_array()[4]),
        i16::from(vector.as_array()[5]),
        i16::from(vector.as_array()[6]),
        i16::from(vector.as_array()[7]),
])

Avx2

• _mm_cvtepi8_epi16

  • PMOVSXBW xmm, xmm

impl ExtendingCast<I8x16> for I32x4

fn extending_cast_from(vector: I8x16) -> I32x4

Scalar Equivalent:

I32x4::from([
        i32::from(vector.as_array()[0]),
        i32::from(vector.as_array()[1]),
        i32::from(vector.as_array()[2]),
        i32::from(vector.as_array()[3]),
])

Avx2

• _mm_cvtepi8_epi32

  • PMOVSXBD xmm, xmm

impl ExtendingCast<I8x16> for I32x8

fn extending_cast_from(vector: I8x16) -> I32x8

Scalar Equivalent:

I32x8::from([
        i32::from(vector.as_array()[0]),
        i32::from(vector.as_array()[1]),
        i32::from(vector.as_array()[2]),
        i32::from(vector.as_array()[3]),
        i32::from(vector.as_array()[4]),
        i32::from(vector.as_array()[5]),
        i32::from(vector.as_array()[6]),
        i32::from(vector.as_array()[7]),
])

Avx2

• _mm256_cvtepi8_epi32

  • VPMOVSXBD ymm, xmm

impl ExtendingCast<I8x16> for I64x2

fn extending_cast_from(vector: I8x16) -> I64x2

Scalar Equivalent:

I64x2::from([
        i64::from(vector.as_array()[0]),
        i64::from(vector.as_array()[1]),
])

Avx2

• _mm_cvtepi8_epi64

  • PMOVSXBQ xmm, xmm

impl ExtendingCast<I8x16> for I64x4

fn extending_cast_from(vector: I8x16) -> I64x4

Scalar Equivalent:

I64x4::from([
        i64::from(vector.as_array()[0]),
        i64::from(vector.as_array()[1]),
        i64::from(vector.as_array()[2]),
        i64::from(vector.as_array()[3]),
])

Avx2

• _mm256_cvtepi8_epi64

  • VPMOVSXBQ ymm, xmm

impl From<[i8; 16]> for I8x16

fn from(arr: [i8; 16]) -> I8x16

Converts to this type from the input type.

impl From<I16x8> for I8x16

fn from(x: I16x8) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of I16x8 as little endian bits of I8x16.

impl From<I32x4> for I8x16

fn from(x: I32x4) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of I32x4 as little endian bits of I8x16.

impl From<I64x2> for I8x16

fn from(x: I64x2) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of I64x2 as little endian bits of I8x16.

impl From<I8x16> for [i8; 16]

fn from(arr: I8x16) -> [i8; 16]

Converts to this type from the input type.

impl From<I8x16> for I16x16

fn from(vector: I8x16) -> I16x16

Scalar Equivalent:

I16x16::from([
        i16::from(vector.as_array()[0]),
        i16::from(vector.as_array()[1]),
        i16::from(vector.as_array()[2]),
        i16::from(vector.as_array()[3]),
        i16::from(vector.as_array()[4]),
        i16::from(vector.as_array()[5]),
        i16::from(vector.as_array()[6]),
        i16::from(vector.as_array()[7]),
        i16::from(vector.as_array()[8]),
        i16::from(vector.as_array()[9]),
        i16::from(vector.as_array()[10]),
        i16::from(vector.as_array()[11]),
        i16::from(vector.as_array()[12]),
        i16::from(vector.as_array()[13]),
        i16::from(vector.as_array()[14]),
        i16::from(vector.as_array()[15]),
])

Avx2

• _mm256_cvtepi8_epi16

  • VPMOVSXBW ymm, xmm

impl From<I8x16> for I16x8

fn from(x: I8x16) -> I16x8

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of I16x8.

impl From<I8x16> for I32x4

fn from(x: I8x16) -> I32x4

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of I32x4.

impl From<I8x16> for I64x2

fn from(x: I8x16) -> I64x2

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of I64x2.

impl From<I8x16> for I8x32

fn from(vector: I8x16) -> I8x32

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; 32];
out[0..16].copy_from_slice(&vector.as_array());
I8x32::from(out)

Avx2

• _mm256_zextsi128_si256

impl From<I8x16> for U16x8

fn from(x: I8x16) -> U16x8

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of U16x8.

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<I8x16> for U64x2

fn from(x: I8x16) -> U64x2

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of U64x2.

impl From<I8x16> for U8x16

fn from(x: I8x16) -> U8x16

This cast is 100% free. It reinterprets the little-endinan bits of I8x16 as little endian bits of U8x16.

impl From<U16x8> for I8x16

fn from(x: U16x8) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of U16x8 as little endian bits of I8x16.

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<U64x2> for I8x16

fn from(x: U64x2) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of U64x2 as little endian bits of I8x16.

impl From<U8x16> for I8x16

fn from(x: U8x16) -> I8x16

This cast is 100% free. It reinterprets the little-endinan bits of U8x16 as little endian bits of I8x16.

impl Hash for I8x16

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 I8x16

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 I8x16

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 I8x16

fn shl(self, amount: u64) -> I8x16

Scalar Equivalent:

if amount >= 8 {
    I8x16::ZERO
} else {
    I8x16::from([
        self.as_array()[0] << amount,
        self.as_array()[1] << amount,
        self.as_array()[2] << amount,
        self.as_array()[3] << amount,
        self.as_array()[4] << amount,
        self.as_array()[5] << amount,
        self.as_array()[6] << amount,
        self.as_array()[7] << amount,
        self.as_array()[8] << amount,
        self.as_array()[9] << amount,
        self.as_array()[10] << amount,
        self.as_array()[11] << amount,
        self.as_array()[12] << amount,
        self.as_array()[13] << amount,
        self.as_array()[14] << amount,
        self.as_array()[15] << amount,
    ])
}

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

type Output = I8x16

The resulting type after applying the << operator.

impl Shl for I8x16

fn shl(self, amount: I8x16) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for (x, amm) in out.iter_mut().zip(amount.as_array().iter().copied()) {
    *x = if (0..8).contains(&amm) {
        *x << amm
    }  else {
        0
    };
}
I8x16::from(out)

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

type Output = I8x16

The resulting type after applying the << operator.

impl ShlAssign<u64> for I8x16

fn shl_assign(&mut self, amount: u64)

Performs the <<= operation.

impl ShlAssign for I8x16

fn shl_assign(&mut self, amount: I8x16)

Performs the <<= operation.

impl Shr<u64> for I8x16

fn shr(self, amount: u64) -> I8x16

Scalar Equivalent:

if amount >= 8 {
    let mut out = self.as_array();
    for x in out.iter_mut() {
        *x = if *x < 0 { -1 } else { 0 };
    }
    I8x16::from(out)
} else {
    I8x16::from([
        self.as_array()[0] >> amount,
        self.as_array()[1] >> amount,
        self.as_array()[2] >> amount,
        self.as_array()[3] >> amount,
        self.as_array()[4] >> amount,
        self.as_array()[5] >> amount,
        self.as_array()[6] >> amount,
        self.as_array()[7] >> amount,
        self.as_array()[8] >> amount,
        self.as_array()[9] >> amount,
        self.as_array()[10] >> amount,
        self.as_array()[11] >> amount,
        self.as_array()[12] >> amount,
        self.as_array()[13] >> amount,
        self.as_array()[14] >> amount,
        self.as_array()[15] >> amount,
    ])
}

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

type Output = I8x16

The resulting type after applying the >> operator.

impl Shr for I8x16

fn shr(self, amount: I8x16) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for (x, amm) in out.iter_mut().zip(amount.as_array().iter().copied()) {
    *x = if (0..8).contains(&amm) {
        *x >> amm
    } else if *x < 0 {
        -1
    }  else {
        0
    };
}
I8x16::from(out)

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

type Output = I8x16

The resulting type after applying the >> operator.

impl ShrAssign<u64> for I8x16

fn shr_assign(&mut self, amount: u64)

Performs the >>= operation.

impl ShrAssign for I8x16

fn shr_assign(&mut self, amount: I8x16)

Performs the >>= operation.

impl SimdBase for I8x16

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: i8) -> I8x16

Scalar Equivalent:

let mut out = [0; 16];
out[0] = scalar;
I8x16::from(out)

Avx2

• _mm_set_epi8

  Instruction sequence.

fn extract<const I: usize>(&self) -> i8

Scalar Equivalent:

self.as_array()[I]

Avx2

• _mm_extract_epi8

  • PEXTRB r32, xmm, imm8

fn broadcast(scalar: i8) -> I8x16

Scalar Equivalent:

I8x16::from([scalar; 16])

Avx2

• _mm_set1_epi8

  Instruction sequence.

fn broadcast_lo(vector: I8x16) -> I8x16

Scalar Equivalent:

I8x16::from([vector.as_array()[0]; 16])

Avx2

• _mm_broadcastb_epi8

  • VPBROADCASTB xmm, xmm

fn cmp_eq(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::from([
    if self.as_array()[0] == other.as_array()[0] {  -1  } else { 0 },
    if self.as_array()[1] == other.as_array()[1] {  -1  } else { 0 },
    if self.as_array()[2] == other.as_array()[2] {  -1  } else { 0 },
    if self.as_array()[3] == other.as_array()[3] {  -1  } else { 0 },
    if self.as_array()[4] == other.as_array()[4] {  -1  } else { 0 },
    if self.as_array()[5] == other.as_array()[5] {  -1  } else { 0 },
    if self.as_array()[6] == other.as_array()[6] {  -1  } else { 0 },
    if self.as_array()[7] == other.as_array()[7] {  -1  } else { 0 },
    if self.as_array()[8] == other.as_array()[8] {  -1  } else { 0 },
    if self.as_array()[9] == other.as_array()[9] {  -1  } else { 0 },
    if self.as_array()[10] == other.as_array()[10] {  -1  } else { 0 },
    if self.as_array()[11] == other.as_array()[11] {  -1  } else { 0 },
    if self.as_array()[12] == other.as_array()[12] {  -1  } else { 0 },
    if self.as_array()[13] == other.as_array()[13] {  -1  } else { 0 },
    if self.as_array()[14] == other.as_array()[14] {  -1  } else { 0 },
    if self.as_array()[15] == other.as_array()[15] {  -1  } else { 0 },
])

Avx2

• _mm_cmpeq_epi8

  • PCMPEQB xmm, xmm

fn and_not(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::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]),
    self.as_array()[4] & (!other.as_array()[4]),
    self.as_array()[5] & (!other.as_array()[5]),
    self.as_array()[6] & (!other.as_array()[6]),
    self.as_array()[7] & (!other.as_array()[7]),
    self.as_array()[8] & (!other.as_array()[8]),
    self.as_array()[9] & (!other.as_array()[9]),
    self.as_array()[10] & (!other.as_array()[10]),
    self.as_array()[11] & (!other.as_array()[11]),
    self.as_array()[12] & (!other.as_array()[12]),
    self.as_array()[13] & (!other.as_array()[13]),
    self.as_array()[14] & (!other.as_array()[14]),
    self.as_array()[15] & (!other.as_array()[15]),
])

Avx2

• _mm_andnot_si128

  • PANDN xmm, xmm

fn cmp_gt(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::from([
    if self.as_array()[0] > other.as_array()[0] {  -1  } else { 0 },
    if self.as_array()[1] > other.as_array()[1] {  -1  } else { 0 },
    if self.as_array()[2] > other.as_array()[2] {  -1  } else { 0 },
    if self.as_array()[3] > other.as_array()[3] {  -1  } else { 0 },
    if self.as_array()[4] > other.as_array()[4] {  -1  } else { 0 },
    if self.as_array()[5] > other.as_array()[5] {  -1  } else { 0 },
    if self.as_array()[6] > other.as_array()[6] {  -1  } else { 0 },
    if self.as_array()[7] > other.as_array()[7] {  -1  } else { 0 },
    if self.as_array()[8] > other.as_array()[8] {  -1  } else { 0 },
    if self.as_array()[9] > other.as_array()[9] {  -1  } else { 0 },
    if self.as_array()[10] > other.as_array()[10] {  -1  } else { 0 },
    if self.as_array()[11] > other.as_array()[11] {  -1  } else { 0 },
    if self.as_array()[12] > other.as_array()[12] {  -1  } else { 0 },
    if self.as_array()[13] > other.as_array()[13] {  -1  } else { 0 },
    if self.as_array()[14] > other.as_array()[14] {  -1  } else { 0 },
    if self.as_array()[15] > other.as_array()[15] {  -1  } else { 0 },
])

Avx2

• _mm_cmpgt_epi8

  • PCMPGTB xmm, xmm

fn shift_left<const BITS: usize>(&self) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for x in out.iter_mut() {
    *x <<= BITS;
}
I8x16::from(out)

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

fn shift_right<const BITS: usize>(&self) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for x in out.iter_mut() {
    *x >>= BITS;
}
I8x16::from(out)

Avx2

WARNING: this implementation is a polyfill which executes the scalar implemenation.

fn unpack_lo(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::from([
    // Lane# 0
    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],
    self.as_array()[4],
    other.as_array()[4],
    self.as_array()[5],
    other.as_array()[5],
    self.as_array()[6],
    other.as_array()[6],
    self.as_array()[7],
    other.as_array()[7],
])

Avx2

• _mm_unpacklo_epi8

  • PUNPCKLBW xmm, xmm

fn unpack_hi(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::from([
    // Lane# 0
    self.as_array()[8],
    other.as_array()[8],
    self.as_array()[9],
    other.as_array()[9],
    self.as_array()[10],
    other.as_array()[10],
    self.as_array()[11],
    other.as_array()[11],
    self.as_array()[12],
    other.as_array()[12],
    self.as_array()[13],
    other.as_array()[13],
    self.as_array()[14],
    other.as_array()[14],
    self.as_array()[15],
    other.as_array()[15],
])

Avx2

• _mm_unpackhi_epi8

  • PUNPCKHBW xmm, xmm

fn max(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::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]),
    self.as_array()[4].max(other.as_array()[4]),
    self.as_array()[5].max(other.as_array()[5]),
    self.as_array()[6].max(other.as_array()[6]),
    self.as_array()[7].max(other.as_array()[7]),
    self.as_array()[8].max(other.as_array()[8]),
    self.as_array()[9].max(other.as_array()[9]),
    self.as_array()[10].max(other.as_array()[10]),
    self.as_array()[11].max(other.as_array()[11]),
    self.as_array()[12].max(other.as_array()[12]),
    self.as_array()[13].max(other.as_array()[13]),
    self.as_array()[14].max(other.as_array()[14]),
    self.as_array()[15].max(other.as_array()[15]),
])

Avx2

• _mm_max_epi8

  • PMAXSB xmm, xmm

fn min(&self, other: I8x16) -> I8x16

Scalar Equivalent:

I8x16::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]),
    self.as_array()[4].min(other.as_array()[4]),
    self.as_array()[5].min(other.as_array()[5]),
    self.as_array()[6].min(other.as_array()[6]),
    self.as_array()[7].min(other.as_array()[7]),
    self.as_array()[8].min(other.as_array()[8]),
    self.as_array()[9].min(other.as_array()[9]),
    self.as_array()[10].min(other.as_array()[10]),
    self.as_array()[11].min(other.as_array()[11]),
    self.as_array()[12].min(other.as_array()[12]),
    self.as_array()[13].min(other.as_array()[13]),
    self.as_array()[14].min(other.as_array()[14]),
    self.as_array()[15].min(other.as_array()[15]),
])

Avx2

• _mm_min_epi8

  • PMINSB xmm, xmm

const LANES: usize = 16usize

The number of elements of this vector.

const ZERO: Self

A vector where every element is zero.

type Scalar = i8

The scalar that this value holds.

type Array = [i8; 16]

The equivalent array type of this vector.

type Signed = I8x16

The signed version of this vector.

type Unsigned = U8x16

The unsigned version of this vector.

type BroadcastLoInput = I8x16

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 SimdBase8 for I8x16

fn shift_bytes_left<const AMOUNT: usize>(&self) -> I8x16

Scalar Equivalent:

let mut out = [0; 16];
for (out_lane, src_lane) in out
    .chunks_exact_mut(16)
    .zip(self.as_array().chunks_exact(16))
{
    out_lane[AMOUNT..].copy_from_slice(&src_lane[0..16 - AMOUNT]);
}
I8x16::from(out)

Avx2

• _mm_slli_si128

  • PSLLDQ xmm, imm8

fn shift_bytes_right<const AMOUNT: usize>(&self) -> I8x16

Scalar Equivalent:

let mut out = [0; 16];
for (out_lane, src_lane) in out
    .chunks_exact_mut(16)
    .zip(self.as_array().chunks_exact(16))
{
    out_lane[0..16 - AMOUNT].copy_from_slice(&src_lane[AMOUNT..]);
}
I8x16::from(out)

Avx2

• _mm_srli_si128

  • PSRLDQ xmm, imm8

fn most_significant_bits(&self) -> u32

Scalar Equivalent:

let mut out: u32 = 0;
for (i, value) in self.as_array().iter().copied().enumerate() {
    out |= u32::from((value as u8) >> 7) << i;
}
out

Avx2

• _mm_movemask_epi8

  • PMOVMSKB r32, xmm

impl SimdSaturatingArithmetic for I8x16

fn saturating_add(&self, other: I8x16) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for (dst, src) in out.iter_mut().zip(other.as_array().iter()) {
    *dst = dst.saturating_add(*src);
}
Self::from(out)

Avx2

• _mm_adds_epi8

  • PADDSB xmm, xmm

fn saturating_sub(&self, other: I8x16) -> I8x16

Scalar Equivalent:

let mut out = self.as_array();
for (dst, src) in out.iter_mut().zip(other.as_array().iter()) {
    *dst = dst.saturating_sub(*src);
}
Self::from(out)

Avx2

• _mm_subs_epi8

  • PSUBSB xmm, xmm

impl Sub for I8x16

fn sub(self, rhs: I8x16) -> I8x16

Perform a pairwise wrapping_sub

Scalar Equivalent

I8x16::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]),
    self.as_array()[4].wrapping_sub(rhs.as_array()[4]),
    self.as_array()[5].wrapping_sub(rhs.as_array()[5]),
    self.as_array()[6].wrapping_sub(rhs.as_array()[6]),
    self.as_array()[7].wrapping_sub(rhs.as_array()[7]),
    self.as_array()[8].wrapping_sub(rhs.as_array()[8]),
    self.as_array()[9].wrapping_sub(rhs.as_array()[9]),
    self.as_array()[10].wrapping_sub(rhs.as_array()[10]),
    self.as_array()[11].wrapping_sub(rhs.as_array()[11]),
    self.as_array()[12].wrapping_sub(rhs.as_array()[12]),
    self.as_array()[13].wrapping_sub(rhs.as_array()[13]),
    self.as_array()[14].wrapping_sub(rhs.as_array()[14]),
    self.as_array()[15].wrapping_sub(rhs.as_array()[15]),
])

AVX2 Intrinsics Used

• _mm_sub_epi8
  • PSUBB xmm, xmm

Neon Intrinsics Used

• vsubq_s8
  • This intrinsic compiles to the following instructions:
    • SUB Vd.16B,Vn.16B,Vm.16B

type Output = I8x16

The resulting type after applying the - operator.

impl SubAssign for I8x16

fn sub_assign(&mut self, other: I8x16)

Performs the -= operation.

impl Zeroable for I8x16

fn zeroed() -> Self

Calls zeroed.

impl Copy for I8x16

impl Eq for I8x16

impl Pod for I8x16