Auto merge of #111850 - the8472:external-step-by, r=scottmcm

Specialize `StepBy<Range<{integer}>>`

OLD

    iter::bench_range_step_by_fold_u16      700.00ns/iter +/- 10.00ns
    iter::bench_range_step_by_fold_usize    519.00ns/iter  +/- 6.00ns
    iter::bench_range_step_by_loop_u32      555.00ns/iter  +/- 7.00ns
    iter::bench_range_step_by_sum_reducible  37.00ns/iter  +/- 0.00ns

NEW

    iter::bench_range_step_by_fold_u16       49.00ns/iter +/- 0.00ns
    iter::bench_range_step_by_fold_usize    194.00ns/iter +/- 1.00ns
    iter::bench_range_step_by_loop_u32       98.00ns/iter +/- 0.00ns
    iter::bench_range_step_by_sum_reducible   1.00ns/iter +/- 0.00ns

NEW + `-Ctarget-cpu=x86-64-v3`

    iter::bench_range_step_by_fold_u16      22.00ns/iter +/- 0.00ns
    iter::bench_range_step_by_fold_usize    80.00ns/iter +/- 1.00ns
    iter::bench_range_step_by_loop_u32      41.00ns/iter +/- 0.00ns
    iter::bench_range_step_by_sum_reducible  1.00ns/iter +/- 0.00ns

I have only optimized for walltime of those methods, I haven't tested whether it eliminates bounds checks when indexing into slices via things like `(0..slice.len()).step_by(16)`.

library/core/benches/iter.rs

@@ -2,6 +2,7 @@ use core::borrow::Borrow;
 use core::iter::*;
 use core::mem;
 use core::num::Wrapping;
+use core::ops::Range;
 use test::{black_box, Bencher};
 
 #[bench]
@@ -69,6 +70,57 @@ fn bench_max(b: &mut Bencher) {
     })
 }
 
+#[bench]
+fn bench_range_step_by_sum_reducible(b: &mut Bencher) {
+    let r = 0u32..1024;
+    b.iter(|| {
+        let r = black_box(r.clone()).step_by(8);
+
+        let mut sum: u32 = 0;
+        for i in r {
+            sum += i;
+        }
+
+        sum
+    })
+}
+
+#[bench]
+fn bench_range_step_by_loop_u32(b: &mut Bencher) {
+    let r = 0..(u16::MAX as u32);
+    b.iter(|| {
+        let r = black_box(r.clone()).step_by(64);
+
+        let mut sum: u32 = 0;
+        for i in r {
+            let i = i ^ i.wrapping_sub(1);
+            sum = sum.wrapping_add(i);
+        }
+
+        sum
+    })
+}
+
+#[bench]
+fn bench_range_step_by_fold_usize(b: &mut Bencher) {
+    let r: Range<usize> = 0..(u16::MAX as usize);
+    b.iter(|| {
+        let r = black_box(r.clone());
+        r.step_by(64)
+            .map(|x: usize| x ^ (x.wrapping_sub(1)))
+            .fold(0usize, |acc, i| acc.wrapping_add(i))
+    })
+}
+
+#[bench]
+fn bench_range_step_by_fold_u16(b: &mut Bencher) {
+    let r: Range<u16> = 0..u16::MAX;
+    b.iter(|| {
+        let r = black_box(r.clone());
+        r.step_by(64).map(|x: u16| x ^ (x.wrapping_sub(1))).fold(0u16, |acc, i| acc.wrapping_add(i))
+    })
+}
+
 pub fn copy_zip(xs: &[u8], ys: &mut [u8]) {
     for (a, b) in ys.iter_mut().zip(xs) {
         *a = *b;