merge from upstream

Author: andy-thomason

README.md

@@ -12,6 +12,40 @@ We can also be found on [Zulip][zulip-project-portable-simd].
 
 If you are interested in support for a specific architecture, you may want [stdarch] instead.
 
+## Hello World
+
+Now we're gonna dip our toes into this world with a small SIMD "Hello, World!" example. Make sure your compiler is up to date and using `nightly`. We can do that by running 
+
+```bash
+rustup update -- nightly
+```
+
+or by setting up `rustup default nightly` or else with `cargo +nightly {build,test,run}`. After updating, run 
+```bash
+cargo new hellosimd
+```
+to create a new crate. Edit `hellosimd/Cargo.toml` to be 
+```toml
+[package]
+name = "hellosimd"
+version = "0.1.0"
+edition = "2018"
+[dependencies]
+core_simd = { git = "https://github.com/rust-lang/stdsimd" }
+```
+
+and finally write this in `src/main.rs`:
+```rust
+use core_simd::*;
+fn main() {
+    let a = f32x4::splat(10.0);
+    let b = f32x4::from_array([1.0, 2.0, 3.0, 4.0]);
+    println!("{:?}", a + b);
+}
+```
+
+Explanation: We import all the bindings from the crate with the first line. Then, we construct our SIMD vectors with methods like `splat` or `from_array`. Finally, we can use operators on them like `+` and the appropriate SIMD instructions will be carried out. When we run `cargo run` you should get `[11.0, 12.0, 13.0, 14.0]`.
+
 ## Code Organization
 
 Currently the crate is organized so that each element type is a file, and then the 64-bit, 128-bit, 256-bit, and 512-bit vectors using those types are contained in said file.

crates/core_simd/examples/nbody.rs

@@ -0,0 +1,182 @@
+/// Benchmarks game nbody code
+/// Taken from the `packed_simd` crate
+/// Run this benchmark with `cargo test --example nbody`
+use core_simd::*;
+
+use std::f64::consts::PI;
+const SOLAR_MASS: f64 = 4.0 * PI * PI;
+const DAYS_PER_YEAR: f64 = 365.24;
+
+#[derive(Debug, Clone, Copy)]
+pub struct Body {
+    pub x: f64x4,
+    pub v: f64x4,
+    pub mass: f64,
+}
+
+const N_BODIES: usize = 5;
+const BODIES: [Body; N_BODIES] = [
+    // sun:
+    Body {
+        x: f64x4::from_array([0., 0., 0., 0.]),
+        v: f64x4::from_array([0., 0., 0., 0.]),
+        mass: SOLAR_MASS,
+    },
+    // jupiter:
+    Body {
+        x: f64x4::from_array([
+            4.84143144246472090e+00,
+            -1.16032004402742839e+00,
+            -1.03622044471123109e-01,
+            0.,
+        ]),
+        v: f64x4::from_array([
+            1.66007664274403694e-03 * DAYS_PER_YEAR,
+            7.69901118419740425e-03 * DAYS_PER_YEAR,
+            -6.90460016972063023e-05 * DAYS_PER_YEAR,
+            0.,
+        ]),
+        mass: 9.54791938424326609e-04 * SOLAR_MASS,
+    },
+    // saturn:
+    Body {
+        x: f64x4::from_array([
+            8.34336671824457987e+00,
+            4.12479856412430479e+00,
+            -4.03523417114321381e-01,
+            0.,
+        ]),
+        v: f64x4::from_array([
+            -2.76742510726862411e-03 * DAYS_PER_YEAR,
+            4.99852801234917238e-03 * DAYS_PER_YEAR,
+            2.30417297573763929e-05 * DAYS_PER_YEAR,
+            0.,
+        ]),
+        mass: 2.85885980666130812e-04 * SOLAR_MASS,
+    },
+    // uranus:
+    Body {
+        x: f64x4::from_array([
+            1.28943695621391310e+01,
+            -1.51111514016986312e+01,
+            -2.23307578892655734e-01,
+            0.,
+        ]),
+        v: f64x4::from_array([
+            2.96460137564761618e-03 * DAYS_PER_YEAR,
+            2.37847173959480950e-03 * DAYS_PER_YEAR,
+            -2.96589568540237556e-05 * DAYS_PER_YEAR,
+            0.,
+        ]),
+        mass: 4.36624404335156298e-05 * SOLAR_MASS,
+    },
+    // neptune:
+    Body {
+        x: f64x4::from_array([
+            1.53796971148509165e+01,
+            -2.59193146099879641e+01,
+            1.79258772950371181e-01,
+            0.,
+        ]),
+        v: f64x4::from_array([
+            2.68067772490389322e-03 * DAYS_PER_YEAR,
+            1.62824170038242295e-03 * DAYS_PER_YEAR,
+            -9.51592254519715870e-05 * DAYS_PER_YEAR,
+            0.,
+        ]),
+        mass: 5.15138902046611451e-05 * SOLAR_MASS,
+    },
+];
+
+pub fn offset_momentum(bodies: &mut [Body; N_BODIES]) {
+    let (sun, rest) = bodies.split_at_mut(1);
+    let sun = &mut sun[0];
+    for body in rest {
+        let m_ratio = body.mass / SOLAR_MASS;
+        sun.v -= body.v * m_ratio;
+    }
+}
+
+pub fn energy(bodies: &[Body; N_BODIES]) -> f64 {
+    let mut e = 0.;
+    for i in 0..N_BODIES {
+        let bi = &bodies[i];
+        e += bi.mass * (bi.v * bi.v).horizontal_sum() * 0.5;
+        for bj in bodies.iter().take(N_BODIES).skip(i + 1) {
+            let dx = bi.x - bj.x;
+            e -= bi.mass * bj.mass / (dx * dx).horizontal_sum().sqrt()
+        }
+    }
+    e
+}
+
+pub fn advance(bodies: &mut [Body; N_BODIES], dt: f64) {
+    const N: usize = N_BODIES * (N_BODIES - 1) / 2;
+
+    // compute distance between bodies:
+    let mut r = [f64x4::splat(0.); N];
+    {
+        let mut i = 0;
+        for j in 0..N_BODIES {
+            for k in j + 1..N_BODIES {
+                r[i] = bodies[j].x - bodies[k].x;
+                i += 1;
+            }
+        }
+    }
+
+    let mut mag = [0.0; N];
+    for i in (0..N).step_by(2) {
+        let d2s = f64x2::from_array([
+            (r[i] * r[i]).horizontal_sum(),
+            (r[i + 1] * r[i + 1]).horizontal_sum(),
+        ]);
+        let dmags = f64x2::splat(dt) / (d2s * d2s.sqrt());
+        mag[i] = dmags[0];
+        mag[i + 1] = dmags[1];
+    }
+
+    let mut i = 0;
+    for j in 0..N_BODIES {
+        for k in j + 1..N_BODIES {
+            let f = r[i] * mag[i];
+            bodies[j].v -= f * bodies[k].mass;
+            bodies[k].v += f * bodies[j].mass;
+            i += 1
+        }
+    }
+    for body in bodies {
+        body.x += dt * body.v
+    }
+}
+
+pub fn run(n: usize) -> (f64, f64) {
+    let mut bodies = BODIES;
+    offset_momentum(&mut bodies);
+    let energy_before = energy(&bodies);
+    for _ in 0..n {
+        advance(&mut bodies, 0.01);
+    }
+    let energy_after = energy(&bodies);
+
+    (energy_before, energy_after)
+}
+
+#[cfg(test)]
+mod tests {
+    // Good enough for demonstration purposes, not going for strictness here.
+    fn approx_eq_f64(a: f64, b: f64) -> bool {
+        (a - b).abs() < 0.00001
+    }
+    #[test]
+    fn test() {
+        const OUTPUT: [f64; 2] = [-0.169075164, -0.169087605];
+        let (energy_before, energy_after) = super::run(1000);
+        assert!(approx_eq_f64(energy_before, OUTPUT[0]));
+        assert!(approx_eq_f64(energy_after, OUTPUT[1]));
+    }
+}
+
+fn main() {
+    // empty main to pass CI
+}