RMath: first pass at an SSE implementation.

sub_crates/rmath/src/wide4/fallback.rs

@@ -0,0 +1,439 @@
+use std::ops::{Add, BitAnd, BitOr, BitXor, Div, Index, Mul, Neg, Not, Sub};
+
+use crate::FMulAdd;
+
+//=============================================================
+// Float4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(C, align(16))]
+pub struct Float4([f32; 4]);
+
+impl Float4 {
+    /// Create a new `Float4` with the given components.
+    #[inline(always)]
+    pub fn new(a: f32, b: f32, c: f32, d: f32) -> Self {
+        Self([a, b, c, d])
+    }
+
+    /// Create a new `Float4` with all elements set to `n`.
+    #[inline(always)]
+    pub fn splat(n: f32) -> Self {
+        Self([n, n, n, n])
+    }
+
+    /// Component-wise fused multiply-add.
+    ///
+    /// `(self * a) + b` with only one rounding error.
+    #[inline(always)]
+    pub fn mul_add(self, a: Self, b: Self) -> Self {
+        Self([
+            self.0[0].mul_add(a.0[0], b.0[0]),
+            self.0[1].mul_add(a.0[1], b.0[1]),
+            self.0[2].mul_add(a.0[2], b.0[2]),
+            self.0[3].mul_add(a.0[3], b.0[3]),
+        ])
+    }
+
+    /// Vertical minimum.
+    #[inline(always)]
+    pub fn min(self, a: Self) -> Self {
+        Self([
+            self.0[0].min(a.0[0]),
+            self.0[1].min(a.0[1]),
+            self.0[2].min(a.0[2]),
+            self.0[3].min(a.0[3]),
+        ])
+    }
+
+    /// Vertical maximum.
+    #[inline(always)]
+    pub fn max(self, a: Self) -> Self {
+        Self([
+            self.0[0].max(a.0[0]),
+            self.0[1].max(a.0[1]),
+            self.0[2].max(a.0[2]),
+            self.0[3].max(a.0[3]),
+        ])
+    }
+
+    /// Horizontal minimum.
+    #[inline(always)]
+    pub fn min_element(self) -> f32 {
+        let a = self.0[0].min(self.0[1]);
+        let b = self.0[2].min(self.0[3]);
+        a.min(b)
+    }
+
+    /// Horizontal maximum.
+    #[inline(always)]
+    pub fn max_element(self) -> f32 {
+        let a = self.0[0].max(self.0[1]);
+        let b = self.0[2].max(self.0[3]);
+        a.max(b)
+    }
+
+    /// 1.0 / self
+    #[inline(always)]
+    pub fn recip(self) -> Self {
+        Float4::splat(1.0) / self
+    }
+
+    #[inline(always)]
+    pub fn abs(self) -> Self {
+        Float4::new(
+            self.a().abs(),
+            self.b().abs(),
+            self.c().abs(),
+            self.d().abs(),
+        )
+    }
+
+    //-----------------------------------------------------
+    // Comparisons.
+
+    /// Less than.
+    #[inline(always)]
+    pub fn cmplt(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] < rhs.0[0],
+            self.0[1] < rhs.0[1],
+            self.0[2] < rhs.0[2],
+            self.0[3] < rhs.0[3],
+        ])
+    }
+
+    /// Less than or equal.
+    #[inline(always)]
+    pub fn cmplte(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] <= rhs.0[0],
+            self.0[1] <= rhs.0[1],
+            self.0[2] <= rhs.0[2],
+            self.0[3] <= rhs.0[3],
+        ])
+    }
+
+    /// Greater than.
+    #[inline(always)]
+    pub fn cmpgt(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] > rhs.0[0],
+            self.0[1] > rhs.0[1],
+            self.0[2] > rhs.0[2],
+            self.0[3] > rhs.0[3],
+        ])
+    }
+
+    /// Greater than or equal.
+    #[inline(always)]
+    pub fn cmpgte(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] >= rhs.0[0],
+            self.0[1] >= rhs.0[1],
+            self.0[2] >= rhs.0[2],
+            self.0[3] >= rhs.0[3],
+        ])
+    }
+
+    /// Equal.
+    #[inline(always)]
+    pub fn cmpeq(self, rhs: Self) -> Bool4 {
+        Bool4([
+            self.0[0] == rhs.0[0],
+            self.0[1] == rhs.0[1],
+            self.0[2] == rhs.0[2],
+            self.0[3] == rhs.0[3],
+        ])
+    }
+
+    //-----------------------------------------------------
+    // Individual components.
+
+    #[inline(always)]
+    pub fn a(self) -> f32 {
+        self.0[0]
+    }
+
+    #[inline(always)]
+    pub fn b(self) -> f32 {
+        self.0[1]
+    }
+
+    #[inline(always)]
+    pub fn c(self) -> f32 {
+        self.0[2]
+    }
+
+    #[inline(always)]
+    pub fn d(self) -> f32 {
+        self.0[3]
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_a(self, n: f32) -> Self {
+        Self([n, self.0[1], self.0[2], self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_b(self, n: f32) -> Self {
+        Self([self.0[0], n, self.0[2], self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_c(self, n: f32) -> Self {
+        Self([self.0[0], self.0[1], n, self.0[3]])
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_d(self, n: f32) -> Self {
+        Self([self.0[0], self.0[1], self.0[2], n])
+    }
+
+    //-----------------------------------------------------
+    // Shuffles.
+
+    #[inline(always)]
+    pub fn aaaa(self) -> Self {
+        let a = self.0[0];
+        Self([a, a, a, a])
+    }
+
+    #[inline(always)]
+    pub fn bbbb(self) -> Self {
+        let b = self.0[1];
+        Self([b, b, b, b])
+    }
+
+    #[inline(always)]
+    pub fn cccc(self) -> Self {
+        let c = self.0[2];
+        Self([c, c, c, c])
+    }
+
+    #[inline(always)]
+    pub fn dddd(self) -> Self {
+        let d = self.0[3];
+        Self([d, d, d, d])
+    }
+
+    #[inline(always)]
+    pub fn bcad(self) -> Self {
+        let a = self.0[0];
+        let b = self.0[1];
+        let c = self.0[2];
+        let d = self.0[3];
+        Self([b, c, a, d])
+    }
+
+    #[inline(always)]
+    pub fn cabd(self) -> Self {
+        let a = self.0[0];
+        let b = self.0[1];
+        let c = self.0[2];
+        let d = self.0[3];
+        Self([c, a, b, d])
+    }
+}
+
+impl Index<usize> for Float4 {
+    type Output = f32;
+
+    #[inline(always)]
+    fn index(&self, idx: usize) -> &f32 {
+        &self.0[idx]
+    }
+}
+
+impl Add for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] + rhs.0[0],
+            self.0[1] + rhs.0[1],
+            self.0[2] + rhs.0[2],
+            self.0[3] + rhs.0[3],
+        ])
+    }
+}
+
+impl Sub for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] - rhs.0[0],
+            self.0[1] - rhs.0[1],
+            self.0[2] - rhs.0[2],
+            self.0[3] - rhs.0[3],
+        ])
+    }
+}
+
+impl Mul for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] * rhs.0[0],
+            self.0[1] * rhs.0[1],
+            self.0[2] * rhs.0[2],
+            self.0[3] * rhs.0[3],
+        ])
+    }
+}
+
+impl Mul<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: f32) -> Self {
+        Self([
+            self.0[0] * rhs,
+            self.0[1] * rhs,
+            self.0[2] * rhs,
+            self.0[3] * rhs,
+        ])
+    }
+}
+
+impl Div for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] / rhs.0[0],
+            self.0[1] / rhs.0[1],
+            self.0[2] / rhs.0[2],
+            self.0[3] / rhs.0[3],
+        ])
+    }
+}
+
+impl Div<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: f32) -> Self {
+        Self([
+            self.0[0] / rhs,
+            self.0[1] / rhs,
+            self.0[2] / rhs,
+            self.0[3] / rhs,
+        ])
+    }
+}
+
+impl Neg for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self([-self.0[0], -self.0[1], -self.0[2], -self.0[3]])
+    }
+}
+
+impl FMulAdd for Float4 {
+    #[inline(always)]
+    fn fma(self, b: Self, c: Self) -> Self {
+        self.mul_add(b, c)
+    }
+}
+
+//=============================================================
+// Bool4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Bool4([bool; 4]);
+
+impl Bool4 {
+    #[inline(always)]
+    pub fn new_false() -> Self {
+        Self([false, false, false, false])
+    }
+
+    #[inline(always)]
+    pub fn to_bools(self) -> [bool; 4] {
+        self.0
+    }
+
+    /// Note: `a` goes to the least significant bit.
+    #[inline(always)]
+    pub fn bitmask(self) -> u8 {
+        self.0[0] as u8
+            | ((self.0[1] as u8) << 1)
+            | ((self.0[2] as u8) << 2)
+            | ((self.0[3] as u8) << 3)
+    }
+
+    #[inline(always)]
+    pub fn any(self) -> bool {
+        self.0[0] | &self.0[1] | self.0[2] | self.0[3]
+    }
+
+    #[inline(always)]
+    pub fn all(self) -> bool {
+        self.0[0] & &self.0[1] & self.0[2] & self.0[3]
+    }
+}
+
+impl BitAnd for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitand(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] & rhs.0[0],
+            self.0[1] & rhs.0[1],
+            self.0[2] & rhs.0[2],
+            self.0[3] & rhs.0[3],
+        ])
+    }
+}
+
+impl BitOr for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitor(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] | rhs.0[0],
+            self.0[1] | rhs.0[1],
+            self.0[2] | rhs.0[2],
+            self.0[3] | rhs.0[3],
+        ])
+    }
+}
+
+impl BitXor for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitxor(self, rhs: Self) -> Self {
+        Self([
+            self.0[0] ^ rhs.0[0],
+            self.0[1] ^ rhs.0[1],
+            self.0[2] ^ rhs.0[2],
+            self.0[3] ^ rhs.0[3],
+        ])
+    }
+}
+
+impl Not for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn not(self) -> Self {
+        Self([!self.0[0], !self.0[1], !self.0[2], !self.0[3]])
+    }
+}

sub_crates/rmath/src/wide4/mod.rs

@@ -1,453 +1,25 @@
-use std::ops::{
-    AddAssign, BitAndAssign, BitOrAssign, BitXorAssign, DivAssign, MulAssign, SubAssign,
+use std::{
+    cmp::PartialEq,
+    ops::{AddAssign, BitAndAssign, BitOrAssign, BitXorAssign, DivAssign, MulAssign, SubAssign},
 };
 
-use std::cmp::PartialEq;
-
 use crate::utils::ulps_eq;
 use crate::{difference_of_products, two_prod, two_sum};
 
+//-------------------------------------------------------------
+// Which implementation to use.
+
+#[cfg(not(any(target_arch = "x86_64")))]
+mod fallback;
+#[cfg(not(any(target_arch = "x86_64")))]
 pub use fallback::{Bool4, Float4};
-mod fallback {
-    use std::ops::{Add, BitAnd, BitOr, BitXor, Div, Index, Mul, Neg, Not, Sub};
 
-    use crate::FMulAdd;
-
-    #[derive(Debug, Copy, Clone)]
-    #[repr(C, align(16))]
-    pub struct Float4([f32; 4]);
-
-    impl Float4 {
-        /// Create a new `Float4` with the given components.
-        #[inline(always)]
-        pub fn new(a: f32, b: f32, c: f32, d: f32) -> Self {
-            Self([a, b, c, d])
-        }
-
-        /// Create a new `Float4` with all elements set to `n`.
-        #[inline(always)]
-        pub fn splat(n: f32) -> Self {
-            Self([n, n, n, n])
-        }
-
-        /// Component-wise fused multiply-add.
-        ///
-        /// `(self * a) + b` with only one rounding error.
-        #[inline(always)]
-        pub fn mul_add(self, a: Self, b: Self) -> Self {
-            Self([
-                self.0[0].mul_add(a.0[0], b.0[0]),
-                self.0[1].mul_add(a.0[1], b.0[1]),
-                self.0[2].mul_add(a.0[2], b.0[2]),
-                self.0[3].mul_add(a.0[3], b.0[3]),
-            ])
-        }
-
-        /// Vertical minimum.
-        #[inline(always)]
-        pub fn min(self, a: Self) -> Self {
-            Self([
-                self.0[0].min(a.0[0]),
-                self.0[1].min(a.0[1]),
-                self.0[2].min(a.0[2]),
-                self.0[3].min(a.0[3]),
-            ])
-        }
-
-        /// Vertical maximum.
-        #[inline(always)]
-        pub fn max(self, a: Self) -> Self {
-            Self([
-                self.0[0].max(a.0[0]),
-                self.0[1].max(a.0[1]),
-                self.0[2].max(a.0[2]),
-                self.0[3].max(a.0[3]),
-            ])
-        }
-
-        /// Horizontal minimum.
-        #[inline(always)]
-        pub fn min_element(self) -> f32 {
-            let a = self.0[0].min(self.0[1]);
-            let b = self.0[2].min(self.0[3]);
-            a.min(b)
-        }
-
-        /// Horizontal maximum.
-        #[inline(always)]
-        pub fn max_element(self) -> f32 {
-            let a = self.0[0].max(self.0[1]);
-            let b = self.0[2].max(self.0[3]);
-            a.max(b)
-        }
-
-        /// 1.0 / self
-        #[inline(always)]
-        pub fn recip(self) -> Self {
-            Float4::splat(1.0) / self
-        }
-
-        #[inline(always)]
-        pub fn abs(self) -> Self {
-            Float4::new(
-                self.a().abs(),
-                self.b().abs(),
-                self.c().abs(),
-                self.d().abs(),
-            )
-        }
-
-        //-----------------------------------------------------
-        // Comparisons.
-
-        /// Less than.
-        #[inline(always)]
-        pub fn cmplt(self, rhs: Self) -> Bool4 {
-            Bool4([
-                self.0[0] < rhs.0[0],
-                self.0[1] < rhs.0[1],
-                self.0[2] < rhs.0[2],
-                self.0[3] < rhs.0[3],
-            ])
-        }
-
-        /// Less than or equal.
-        #[inline(always)]
-        pub fn cmplte(self, rhs: Self) -> Bool4 {
-            Bool4([
-                self.0[0] <= rhs.0[0],
-                self.0[1] <= rhs.0[1],
-                self.0[2] <= rhs.0[2],
-                self.0[3] <= rhs.0[3],
-            ])
-        }
-
-        /// Greater than.
-        #[inline(always)]
-        pub fn cmpgt(self, rhs: Self) -> Bool4 {
-            Bool4([
-                self.0[0] > rhs.0[0],
-                self.0[1] > rhs.0[1],
-                self.0[2] > rhs.0[2],
-                self.0[3] > rhs.0[3],
-            ])
-        }
-
-        /// Greater than or equal.
-        #[inline(always)]
-        pub fn cmpgte(self, rhs: Self) -> Bool4 {
-            Bool4([
-                self.0[0] >= rhs.0[0],
-                self.0[1] >= rhs.0[1],
-                self.0[2] >= rhs.0[2],
-                self.0[3] >= rhs.0[3],
-            ])
-        }
-
-        /// Equal.
-        #[inline(always)]
-        pub fn cmpeq(self, rhs: Self) -> Bool4 {
-            Bool4([
-                self.0[0] == rhs.0[0],
-                self.0[1] == rhs.0[1],
-                self.0[2] == rhs.0[2],
-                self.0[3] == rhs.0[3],
-            ])
-        }
-
-        //-----------------------------------------------------
-        // Individual components.
-
-        #[inline(always)]
-        pub fn a(self) -> f32 {
-            self.0[0]
-        }
-
-        #[inline(always)]
-        pub fn b(self) -> f32 {
-            self.0[1]
-        }
-
-        #[inline(always)]
-        pub fn c(self) -> f32 {
-            self.0[2]
-        }
-
-        #[inline(always)]
-        pub fn d(self) -> f32 {
-            self.0[3]
-        }
-
-        #[inline(always)]
-        #[must_use]
-        pub fn set_a(self, n: f32) -> Self {
-            Self([n, self.0[1], self.0[2], self.0[3]])
-        }
-
-        #[inline(always)]
-        #[must_use]
-        pub fn set_b(self, n: f32) -> Self {
-            Self([self.0[0], n, self.0[2], self.0[3]])
-        }
-
-        #[inline(always)]
-        #[must_use]
-        pub fn set_c(self, n: f32) -> Self {
-            Self([self.0[0], self.0[1], n, self.0[3]])
-        }
-
-        #[inline(always)]
-        #[must_use]
-        pub fn set_d(self, n: f32) -> Self {
-            Self([self.0[0], self.0[1], self.0[2], n])
-        }
-
-        //-----------------------------------------------------
-        // Shuffles.
-
-        #[inline(always)]
-        pub fn aaaa(self) -> Self {
-            let a = self.0[0];
-            Self([a, a, a, a])
-        }
-
-        #[inline(always)]
-        pub fn bbbb(self) -> Self {
-            let b = self.0[1];
-            Self([b, b, b, b])
-        }
-
-        #[inline(always)]
-        pub fn cccc(self) -> Self {
-            let c = self.0[2];
-            Self([c, c, c, c])
-        }
-
-        #[inline(always)]
-        pub fn dddd(self) -> Self {
-            let d = self.0[3];
-            Self([d, d, d, d])
-        }
-
-        #[inline(always)]
-        pub fn bcad(self) -> Self {
-            let a = self.0[0];
-            let b = self.0[1];
-            let c = self.0[2];
-            let d = self.0[3];
-            Self([b, c, a, d])
-        }
-
-        #[inline(always)]
-        pub fn cabd(self) -> Self {
-            let a = self.0[0];
-            let b = self.0[1];
-            let c = self.0[2];
-            let d = self.0[3];
-            Self([c, a, b, d])
-        }
-    }
-
-    impl Index<usize> for Float4 {
-        type Output = f32;
-
-        #[inline(always)]
-        fn index(&self, idx: usize) -> &f32 {
-            &self.0[idx]
-        }
-    }
-
-    impl Add for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn add(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] + rhs.0[0],
-                self.0[1] + rhs.0[1],
-                self.0[2] + rhs.0[2],
-                self.0[3] + rhs.0[3],
-            ])
-        }
-    }
-
-    impl Sub for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn sub(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] - rhs.0[0],
-                self.0[1] - rhs.0[1],
-                self.0[2] - rhs.0[2],
-                self.0[3] - rhs.0[3],
-            ])
-        }
-    }
-
-    impl Mul for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn mul(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] * rhs.0[0],
-                self.0[1] * rhs.0[1],
-                self.0[2] * rhs.0[2],
-                self.0[3] * rhs.0[3],
-            ])
-        }
-    }
-
-    impl Mul<f32> for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn mul(self, rhs: f32) -> Self {
-            Self([
-                self.0[0] * rhs,
-                self.0[1] * rhs,
-                self.0[2] * rhs,
-                self.0[3] * rhs,
-            ])
-        }
-    }
-
-    impl Div for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn div(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] / rhs.0[0],
-                self.0[1] / rhs.0[1],
-                self.0[2] / rhs.0[2],
-                self.0[3] / rhs.0[3],
-            ])
-        }
-    }
-
-    impl Div<f32> for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn div(self, rhs: f32) -> Self {
-            Self([
-                self.0[0] / rhs,
-                self.0[1] / rhs,
-                self.0[2] / rhs,
-                self.0[3] / rhs,
-            ])
-        }
-    }
-
-    impl Neg for Float4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn neg(self) -> Self {
-            Self([-self.0[0], -self.0[1], -self.0[2], -self.0[3]])
-        }
-    }
-
-    impl FMulAdd for Float4 {
-        #[inline(always)]
-        fn fma(self, b: Self, c: Self) -> Self {
-            self.mul_add(b, c)
-        }
-    }
-
-    //---------------------------------------------------------
-
-    #[derive(Debug, Copy, Clone)]
-    #[repr(transparent)]
-    pub struct Bool4([bool; 4]);
-
-    impl Bool4 {
-        #[inline(always)]
-        pub fn new_false() -> Self {
-            Self([false, false, false, false])
-        }
-
-        #[inline(always)]
-        pub fn to_bools(self) -> [bool; 4] {
-            self.0
-        }
-
-        /// Note: `a` goes to the least significant bit.
-        #[inline(always)]
-        pub fn bitmask(self) -> u8 {
-            self.0[0] as u8
-                | ((self.0[1] as u8) << 1)
-                | ((self.0[2] as u8) << 2)
-                | ((self.0[3] as u8) << 3)
-        }
-
-        #[inline(always)]
-        pub fn any(self) -> bool {
-            self.0[0] | &self.0[1] | self.0[2] | self.0[3]
-        }
-
-        #[inline(always)]
-        pub fn all(self) -> bool {
-            self.0[0] & &self.0[1] & self.0[2] & self.0[3]
-        }
-    }
-
-    impl BitAnd for Bool4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn bitand(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] & rhs.0[0],
-                self.0[1] & rhs.0[1],
-                self.0[2] & rhs.0[2],
-                self.0[3] & rhs.0[3],
-            ])
-        }
-    }
-
-    impl BitOr for Bool4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn bitor(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] | rhs.0[0],
-                self.0[1] | rhs.0[1],
-                self.0[2] | rhs.0[2],
-                self.0[3] | rhs.0[3],
-            ])
-        }
-    }
-
-    impl BitXor for Bool4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn bitxor(self, rhs: Self) -> Self {
-            Self([
-                self.0[0] ^ rhs.0[0],
-                self.0[1] ^ rhs.0[1],
-                self.0[2] ^ rhs.0[2],
-                self.0[3] ^ rhs.0[3],
-            ])
-        }
-    }
-
-    impl Not for Bool4 {
-        type Output = Self;
-
-        #[inline(always)]
-        fn not(self) -> Self {
-            Self([!self.0[0], !self.0[1], !self.0[2], !self.0[3]])
-        }
-    }
-}
+#[cfg(target_arch = "x86_64")]
+mod sse;
+#[cfg(target_arch = "x86_64")]
+pub use sse::{Bool4, Float4};
 
 //-------------------------------------------------------------
-// Impls that don't depend on inner representation.
 
 impl Float4 {
     /// 3D dot product (only uses the first 3 components).
@@ -838,4 +410,34 @@ mod tests {
         assert!(Float4::aprx_eq(c, d, 575));
         assert!(!Float4::aprx_eq(c, d, 565));
     }
+
+    #[test]
+    fn index() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+        assert_eq!(v[0], 0.0);
+        assert_eq!(v[1], 1.0);
+        assert_eq!(v[2], 2.0);
+        assert_eq!(v[3], 3.0);
+    }
+
+    #[test]
+    fn shuffle() {
+        let v = Float4::new(0.0, 1.0, 2.0, 3.0);
+
+        assert_eq!(v.aaaa(), Float4::splat(0.0));
+        assert_eq!(v.bbbb(), Float4::splat(1.0));
+        assert_eq!(v.cccc(), Float4::splat(2.0));
+        assert_eq!(v.dddd(), Float4::splat(3.0));
+
+        assert_eq!(v.bcad(), Float4::new(1.0, 2.0, 0.0, 3.0));
+        assert_eq!(v.cabd(), Float4::new(2.0, 0.0, 1.0, 3.0));
+    }
+
+    #[test]
+    fn bitmask() {
+        let v1 = Float4::new(0.0, 1.0, 2.0, 3.0);
+        let v2 = Float4::new(9.0, 1.0, 9.0, 3.0);
+
+        assert_eq!(v1.cmpeq(v2).bitmask(), 0b1010);
+    }
 }

sub_crates/rmath/src/wide4/sse.rs

@@ -0,0 +1,370 @@
+use std::ops::{Add, BitAnd, BitOr, BitXor, Div, Index, Mul, Neg, Not, Sub};
+
+use std::arch::x86_64::{
+    __m128, _mm_add_ps, _mm_and_ps, _mm_castsi128_ps, _mm_cmpeq_ps, _mm_cmpge_ps, _mm_cmpgt_ps,
+    _mm_cmple_ps, _mm_cmplt_ps, _mm_div_ps, _mm_fmadd_ps, _mm_max_ps, _mm_min_ps, _mm_movemask_ps,
+    _mm_mul_ps, _mm_or_ps, _mm_rcp_ps, _mm_set1_epi32, _mm_set1_ps, _mm_set_ps, _mm_setzero_ps,
+    _mm_shuffle_ps, _mm_storeu_ps, _mm_sub_ps, _mm_xor_ps,
+};
+
+use crate::FMulAdd;
+
+//=============================================================
+// Float4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Float4(__m128);
+
+impl Float4 {
+    /// Create a new `Float4` with the given components.
+    #[inline(always)]
+    pub fn new(a: f32, b: f32, c: f32, d: f32) -> Self {
+        Self(unsafe { _mm_set_ps(d, c, b, a) })
+    }
+
+    /// Create a new `Float4` with all elements set to `n`.
+    #[inline(always)]
+    pub fn splat(n: f32) -> Self {
+        Self(unsafe { _mm_set1_ps(n) })
+    }
+
+    /// Component-wise fused multiply-add.
+    ///
+    /// `(self * a) + b` with only one rounding error.
+    #[inline(always)]
+    pub fn mul_add(self, a: Self, b: Self) -> Self {
+        if cfg!(feature = "fma") {
+            Self(unsafe { _mm_fmadd_ps(self.0, a.0, b.0) })
+        } else {
+            Self::new(
+                self.a().mul_add(a.a(), b.a()),
+                self.b().mul_add(a.b(), b.b()),
+                self.c().mul_add(a.c(), b.c()),
+                self.d().mul_add(a.d(), b.d()),
+            )
+        }
+    }
+
+    /// Vertical minimum.
+    #[inline(always)]
+    pub fn min(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_min_ps(self.0, rhs.0) })
+    }
+
+    /// Vertical maximum.
+    #[inline(always)]
+    pub fn max(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_max_ps(self.0, rhs.0) })
+    }
+
+    /// Horizontal minimum.
+    #[inline(always)]
+    pub fn min_element(self) -> f32 {
+        let a = self.a().min(self.b());
+        let b = self.c().min(self.d());
+        a.min(b)
+    }
+
+    /// Horizontal maximum.
+    #[inline(always)]
+    pub fn max_element(self) -> f32 {
+        let a = self.a().max(self.b());
+        let b = self.c().max(self.d());
+        a.max(b)
+    }
+
+    /// 1.0 / self
+    #[inline(always)]
+    pub fn recip(self) -> Self {
+        Self(unsafe { _mm_rcp_ps(self.0) })
+    }
+
+    #[inline(always)]
+    pub fn abs(self) -> Self {
+        Self(unsafe {
+            let abs_mask = _mm_castsi128_ps(_mm_set1_epi32(!(1 << 31)));
+            _mm_and_ps(self.0, abs_mask)
+        })
+    }
+
+    //-----------------------------------------------------
+    // Comparisons.
+
+    /// Less than.
+    #[inline(always)]
+    pub fn cmplt(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmplt_ps(self.0, rhs.0) })
+    }
+
+    /// Less than or equal.
+    #[inline(always)]
+    pub fn cmplte(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmple_ps(self.0, rhs.0) })
+    }
+
+    /// Greater than.
+    #[inline(always)]
+    pub fn cmpgt(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpgt_ps(self.0, rhs.0) })
+    }
+
+    /// Greater than or equal.
+    #[inline(always)]
+    pub fn cmpgte(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpge_ps(self.0, rhs.0) })
+    }
+
+    /// Equal.
+    #[inline(always)]
+    pub fn cmpeq(self, rhs: Self) -> Bool4 {
+        Bool4(unsafe { _mm_cmpeq_ps(self.0, rhs.0) })
+    }
+
+    //-----------------------------------------------------
+    // Individual components.
+
+    #[inline(always)]
+    pub fn a(self) -> f32 {
+        self[0]
+    }
+
+    #[inline(always)]
+    pub fn b(self) -> f32 {
+        self[1]
+    }
+
+    #[inline(always)]
+    pub fn c(self) -> f32 {
+        self[2]
+    }
+
+    #[inline(always)]
+    pub fn d(self) -> f32 {
+        self[3]
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_a(self, n: f32) -> Self {
+        Self::new(n, self.b(), self.c(), self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_b(self, n: f32) -> Self {
+        Self::new(self.a(), n, self.c(), self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_c(self, n: f32) -> Self {
+        Self::new(self.a(), self.b(), n, self.d())
+    }
+
+    #[inline(always)]
+    #[must_use]
+    pub fn set_d(self, n: f32) -> Self {
+        Self::new(self.a(), self.b(), self.c(), n)
+    }
+
+    //-----------------------------------------------------
+    // Shuffles.
+
+    #[inline(always)]
+    pub fn aaaa(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b00_00_00_00) })
+    }
+
+    #[inline(always)]
+    pub fn bbbb(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b01_01_01_01) })
+    }
+
+    #[inline(always)]
+    pub fn cccc(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b10_10_10_10) })
+    }
+
+    #[inline(always)]
+    pub fn dddd(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_11_11_11) })
+    }
+
+    #[inline(always)]
+    pub fn bcad(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_00_10_01) })
+    }
+
+    #[inline(always)]
+    pub fn cabd(self) -> Self {
+        Self(unsafe { _mm_shuffle_ps(self.0, self.0, 0b11_01_00_10) })
+    }
+}
+
+impl Index<usize> for Float4 {
+    type Output = f32;
+
+    #[inline(always)]
+    fn index(&self, idx: usize) -> &f32 {
+        let elements: &[f32; 4] = unsafe { std::mem::transmute(&self.0) };
+        match idx {
+            0 => &elements[0],
+            1 => &elements[1],
+            2 => &elements[2],
+            3 => &elements[3],
+            _ => panic!("Out of bounds access of Float4 elements."),
+        }
+    }
+}
+
+impl Add for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn add(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_add_ps(self.0, rhs.0) })
+    }
+}
+
+impl Sub for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn sub(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_sub_ps(self.0, rhs.0) })
+    }
+}
+
+impl Mul for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_mul_ps(self.0, rhs.0) })
+    }
+}
+
+impl Mul<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn mul(self, rhs: f32) -> Self {
+        Self(unsafe { _mm_mul_ps(self.0, _mm_set1_ps(rhs)) })
+    }
+}
+
+impl Div for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_div_ps(self.0, rhs.0) })
+    }
+}
+
+impl Div<f32> for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn div(self, rhs: f32) -> Self {
+        Self(unsafe { _mm_div_ps(self.0, _mm_set1_ps(rhs)) })
+    }
+}
+
+impl Neg for Float4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn neg(self) -> Self {
+        Self(unsafe { _mm_mul_ps(self.0, _mm_set1_ps(-1.0)) })
+    }
+}
+
+impl FMulAdd for Float4 {
+    #[inline(always)]
+    fn fma(self, b: Self, c: Self) -> Self {
+        self.mul_add(b, c)
+    }
+}
+
+//=============================================================
+// Bool4
+
+#[derive(Debug, Copy, Clone)]
+#[repr(transparent)]
+pub struct Bool4(__m128);
+
+impl Bool4 {
+    #[inline(always)]
+    pub fn new_false() -> Self {
+        Self(unsafe { _mm_setzero_ps() })
+    }
+
+    #[inline(always)]
+    pub fn to_bools(self) -> [bool; 4] {
+        let mut v = [0.0f32; 4];
+        unsafe { _mm_storeu_ps((&mut v[..]).as_mut_ptr(), self.0) }
+        [
+            v[0].to_bits() != 0,
+            v[1].to_bits() != 0,
+            v[2].to_bits() != 0,
+            v[3].to_bits() != 0,
+        ]
+    }
+
+    /// Note: `a` goes to the least significant bit.
+    #[inline(always)]
+    pub fn bitmask(self) -> u8 {
+        unsafe { _mm_movemask_ps(self.0) as u8 }
+    }
+
+    #[inline(always)]
+    pub fn any(self) -> bool {
+        self.bitmask() != 0
+    }
+
+    #[inline(always)]
+    pub fn all(self) -> bool {
+        self.bitmask() == 0b1111
+    }
+}
+
+impl BitAnd for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitand(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_and_ps(self.0, rhs.0) })
+    }
+}
+
+impl BitOr for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitor(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_or_ps(self.0, rhs.0) })
+    }
+}
+
+impl BitXor for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn bitxor(self, rhs: Self) -> Self {
+        Self(unsafe { _mm_xor_ps(self.0, rhs.0) })
+    }
+}
+
+impl Not for Bool4 {
+    type Output = Self;
+
+    #[inline(always)]
+    fn not(self) -> Self {
+        Self(unsafe {
+            let ones = _mm_castsi128_ps(_mm_set1_epi32(!0));
+            _mm_xor_ps(self.0, ones)
+        })
+    }
+}