From 8740d5d171fbb2fc9a12e46ecda32a0124badb46 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 20:08:32 +0100
Subject: [PATCH 01/11] Add benchmarks for [u8]::make_ascii_uppercase

---
 src/libcore/benches/ascii_case.rs | 172 ++++++++++++++++++++++++++++++
 src/libcore/benches/lib.rs        |   1 +
 2 files changed, 173 insertions(+)
 create mode 100644 src/libcore/benches/ascii_case.rs

diff --git a/src/libcore/benches/ascii_case.rs b/src/libcore/benches/ascii_case.rs
new file mode 100644
index 0000000000000..43242ca3f364e
--- /dev/null
+++ b/src/libcore/benches/ascii_case.rs
@@ -0,0 +1,172 @@
+// Lower-case ASCII 'a' is the first byte that has its highest bit set after wrap-adding 0x1F:
+//
+//     b'a' + 0x1F == 0x80 == 0b1000_0000
+//     b'z' + 0x1F == 0x98 == 0b10011000
+//
+// Lower-case ASCII 'z' is the last byte that has its highest bit unset after wrap-adding 0x05:
+//
+//     b'a' + 0x05 == 0x66 == 0b0110_0110
+//     b'z' + 0x05 == 0x7F == 0b0111_1111
+//
+// … except for 0xFB to 0xFF, but those are in the range of bytes that have the highest bit
+// unset again after adding 0x1F.
+//
+// So `(byte + 0x1f) & !(byte + 5)` has its highest bit set
+// iff `byte` is a lower-case ASCII letter.
+//
+// Lower-case ASCII letters all have the 0x20 bit set.
+// (Two positions right of 0x80, the highest bit.)
+// Unsetting that bit produces the same letter, in upper-case.
+//
+// Therefore:
+fn branchless_to_ascii_upper_case(byte: u8) -> u8 {
+    byte &
+    !(
+        (
+            byte.wrapping_add(0x1f) &
+            !byte.wrapping_add(0x05) &
+            0x80
+        ) >> 2
+    )
+}
+
+
+macro_rules! benches {
+    ($( fn $name: ident($arg: ident: &mut [u8]) $body: block )+) => {
+        benches!(mod short SHORT $($name $arg $body)+);
+        benches!(mod medium MEDIUM $($name $arg $body)+);
+        benches!(mod long LONG $($name $arg $body)+);
+    };
+
+    (mod $mod_name: ident $input: ident $($name: ident $arg: ident $body: block)+) => {
+        mod $mod_name {
+            use super::*;
+
+            $(
+                #[bench]
+                fn $name(bencher: &mut Bencher) {
+                    bencher.bytes = $input.len() as u64;
+                    bencher.iter(|| {
+                        let mut vec = $input.as_bytes().to_vec();
+                        {
+                            let $arg = &mut vec[..];
+                            $body
+                        }
+                        vec
+                    })
+                }
+            )+
+        }
+    }
+}
+
+use test::black_box;
+use test::Bencher;
+
+benches! {
+    fn bench00_alloc_only(_bytes: &mut [u8]) {}
+
+    fn bench01_black_box_read_each_byte(bytes: &mut [u8]) {
+        for byte in bytes {
+            black_box(*byte);
+        }
+    }
+
+    fn bench02_lookup(bytes: &mut [u8]) {
+        for byte in bytes {
+            *byte = ASCII_UPPERCASE_MAP[*byte as usize]
+        }
+    }
+
+    fn bench03_branch_and_subtract(bytes: &mut [u8]) {
+        for byte in bytes {
+            *byte = if b'a' <= *byte && *byte <= b'z' {
+                *byte - b'a' + b'A'
+            } else {
+                *byte
+            }
+        }
+    }
+
+    fn bench04_branch_and_mask(bytes: &mut [u8]) {
+        for byte in bytes {
+            *byte = if b'a' <= *byte && *byte <= b'z' {
+                *byte & !0x20
+            } else {
+                *byte
+            }
+        }
+    }
+
+    fn bench05_branchless(bytes: &mut [u8]) {
+        for byte in bytes {
+            *byte = branchless_to_ascii_upper_case(*byte)
+        }
+    }
+
+    fn bench06_libcore(bytes: &mut [u8]) {
+        bytes.make_ascii_uppercase()
+    }
+}
+
+macro_rules! repeat {
+    ($s: expr) => { concat!($s, $s, $s, $s, $s, $s, $s, $s, $s, $s) }
+}
+
+const SHORT: &'static str = "Alice's";
+const MEDIUM: &'static str = "Alice's Adventures in Wonderland";
+const LONG: &'static str = repeat!(r#"
+    La Guida di Bragia, a Ballad Opera for the Marionette Theatre (around 1850)
+    Alice's Adventures in Wonderland (1865)
+    Phantasmagoria and Other Poems (1869)
+    Through the Looking-Glass, and What Alice Found There (includes "Jabberwocky" and "The Walrus and the Carpenter") (1871)
+    The Hunting of the Snark (1876)
+    Rhyme? And Reason? (1883) – shares some contents with the 1869 collection, including the long poem "Phantasmagoria"
+    A Tangled Tale (1885)
+    Sylvie and Bruno (1889)
+    Sylvie and Bruno Concluded (1893)
+    Pillow Problems (1893)
+    What the Tortoise Said to Achilles (1895)
+    Three Sunsets and Other Poems (1898)
+    The Manlet (1903)[106]
+"#);
+
+const ASCII_UPPERCASE_MAP: [u8; 256] = [
+    0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+    0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
+    0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
+    0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
+    b' ', b'!', b'"', b'#', b'$', b'%', b'&', b'\'',
+    b'(', b')', b'*', b'+', b',', b'-', b'.', b'/',
+    b'0', b'1', b'2', b'3', b'4', b'5', b'6', b'7',
+    b'8', b'9', b':', b';', b'<', b'=', b'>', b'?',
+    b'@', b'A', b'B', b'C', b'D', b'E', b'F', b'G',
+    b'H', b'I', b'J', b'K', b'L', b'M', b'N', b'O',
+    b'P', b'Q', b'R', b'S', b'T', b'U', b'V', b'W',
+    b'X', b'Y', b'Z', b'[', b'\\', b']', b'^', b'_',
+    b'`',
+
+          b'A', b'B', b'C', b'D', b'E', b'F', b'G',
+    b'H', b'I', b'J', b'K', b'L', b'M', b'N', b'O',
+    b'P', b'Q', b'R', b'S', b'T', b'U', b'V', b'W',
+    b'X', b'Y', b'Z',
+
+                      b'{', b'|', b'}', b'~', 0x7f,
+    0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
+    0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,
+    0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
+    0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
+    0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
+    0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
+    0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
+    0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
+    0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
+    0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
+    0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
+    0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
+    0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
+    0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,
+    0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
+    0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
+];
+
diff --git a/src/libcore/benches/lib.rs b/src/libcore/benches/lib.rs
index 48572af611a5b..b72d035b0ce94 100644
--- a/src/libcore/benches/lib.rs
+++ b/src/libcore/benches/lib.rs
@@ -5,6 +5,7 @@ extern crate core;
 extern crate test;
 
 mod any;
+mod ascii_case;
 mod char;
 mod hash;
 mod iter;

From ce933f77c865a15670855ac5941fe200752b739f Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 20:16:37 +0100
Subject: [PATCH 02/11] Make u8::to_ascii_lowercase and to_ascii_uppercase
 branchless

---
 src/libcore/benches/ascii_case.rs |  22 +-----
 src/libcore/num/mod.rs            | 122 ++++++++++--------------------
 2 files changed, 43 insertions(+), 101 deletions(-)

diff --git a/src/libcore/benches/ascii_case.rs b/src/libcore/benches/ascii_case.rs
index 43242ca3f364e..1c9691d1ee173 100644
--- a/src/libcore/benches/ascii_case.rs
+++ b/src/libcore/benches/ascii_case.rs
@@ -1,24 +1,4 @@
-// Lower-case ASCII 'a' is the first byte that has its highest bit set after wrap-adding 0x1F:
-//
-//     b'a' + 0x1F == 0x80 == 0b1000_0000
-//     b'z' + 0x1F == 0x98 == 0b10011000
-//
-// Lower-case ASCII 'z' is the last byte that has its highest bit unset after wrap-adding 0x05:
-//
-//     b'a' + 0x05 == 0x66 == 0b0110_0110
-//     b'z' + 0x05 == 0x7F == 0b0111_1111
-//
-// … except for 0xFB to 0xFF, but those are in the range of bytes that have the highest bit
-// unset again after adding 0x1F.
-//
-// So `(byte + 0x1f) & !(byte + 5)` has its highest bit set
-// iff `byte` is a lower-case ASCII letter.
-//
-// Lower-case ASCII letters all have the 0x20 bit set.
-// (Two positions right of 0x80, the highest bit.)
-// Unsetting that bit produces the same letter, in upper-case.
-//
-// Therefore:
+// See comments in `u8::to_ascii_uppercase` in `src/libcore/num/mod.rs`.
 fn branchless_to_ascii_upper_case(byte: u8) -> u8 {
     byte &
     !(
diff --git a/src/libcore/num/mod.rs b/src/libcore/num/mod.rs
index d93cfbc2a28ac..977e21e0dc89a 100644
--- a/src/libcore/num/mod.rs
+++ b/src/libcore/num/mod.rs
@@ -3794,7 +3794,39 @@ impl u8 {
     #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
     #[inline]
     pub fn to_ascii_uppercase(&self) -> u8 {
-        ASCII_UPPERCASE_MAP[*self as usize]
+        // See benchmarks in src/libcore/benches/ascii_case.rs
+
+        // Lower-case ASCII 'a' is the first byte that has its highest bit set
+        // after wrap-adding 0x1F:
+        //
+        //     b'a' + 0x1F == 0x80 == 0b1000_0000
+        //     b'z' + 0x1F == 0x98 == 0b10011000
+        //
+        // Lower-case ASCII 'z' is the last byte that has its highest bit unset
+        // after wrap-adding 0x05:
+        //
+        //     b'a' + 0x05 == 0x66 == 0b0110_0110
+        //     b'z' + 0x05 == 0x7F == 0b0111_1111
+        //
+        // … except for 0xFB to 0xFF, but those are in the range of bytes
+        // that have the highest bit unset again after adding 0x1F.
+        //
+        // So `(byte + 0x1f) & !(byte + 5)` has its highest bit set
+        // iff `byte` is a lower-case ASCII letter.
+        //
+        // Lower-case ASCII letters all have the 0x20 bit set.
+        // (Two positions right of 0x80, the highest bit.)
+        // Unsetting that bit produces the same letter, in upper-case.
+        //
+        // Therefore:
+        *self &
+        !(
+            (
+                self.wrapping_add(0x1f) &
+                !self.wrapping_add(0x05) &
+                0x80
+            ) >> 2
+        )
     }
 
     /// Makes a copy of the value in its ASCII lower case equivalent.
@@ -3816,7 +3848,15 @@ impl u8 {
     #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
     #[inline]
     pub fn to_ascii_lowercase(&self) -> u8 {
-        ASCII_LOWERCASE_MAP[*self as usize]
+        // See comments in to_ascii_uppercase above.
+        *self |
+        (
+            (
+                self.wrapping_add(0x3f) &
+                !self.wrapping_add(0x25) &
+                0x80
+            ) >> 2
+        )
     }
 
     /// Checks that two values are an ASCII case-insensitive match.
@@ -4940,84 +4980,6 @@ impl_from! { u32, f64, #[stable(feature = "lossless_float_conv", since = "1.6.0"
 // Float -> Float
 impl_from! { f32, f64, #[stable(feature = "lossless_float_conv", since = "1.6.0")] }
 
-static ASCII_LOWERCASE_MAP: [u8; 256] = [
-    0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
-    0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
-    0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
-    0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
-    b' ', b'!', b'"', b'#', b'$', b'%', b'&', b'\'',
-    b'(', b')', b'*', b'+', b',', b'-', b'.', b'/',
-    b'0', b'1', b'2', b'3', b'4', b'5', b'6', b'7',
-    b'8', b'9', b':', b';', b'<', b'=', b'>', b'?',
-    b'@',
-
-          b'a', b'b', b'c', b'd', b'e', b'f', b'g',
-    b'h', b'i', b'j', b'k', b'l', b'm', b'n', b'o',
-    b'p', b'q', b'r', b's', b't', b'u', b'v', b'w',
-    b'x', b'y', b'z',
-
-                      b'[', b'\\', b']', b'^', b'_',
-    b'`', b'a', b'b', b'c', b'd', b'e', b'f', b'g',
-    b'h', b'i', b'j', b'k', b'l', b'm', b'n', b'o',
-    b'p', b'q', b'r', b's', b't', b'u', b'v', b'w',
-    b'x', b'y', b'z', b'{', b'|', b'}', b'~', 0x7f,
-    0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
-    0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,
-    0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
-    0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
-    0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
-    0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
-    0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
-    0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
-    0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
-    0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
-    0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
-    0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
-    0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
-    0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,
-    0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
-    0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
-];
-
-static ASCII_UPPERCASE_MAP: [u8; 256] = [
-    0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
-    0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
-    0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
-    0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
-    b' ', b'!', b'"', b'#', b'$', b'%', b'&', b'\'',
-    b'(', b')', b'*', b'+', b',', b'-', b'.', b'/',
-    b'0', b'1', b'2', b'3', b'4', b'5', b'6', b'7',
-    b'8', b'9', b':', b';', b'<', b'=', b'>', b'?',
-    b'@', b'A', b'B', b'C', b'D', b'E', b'F', b'G',
-    b'H', b'I', b'J', b'K', b'L', b'M', b'N', b'O',
-    b'P', b'Q', b'R', b'S', b'T', b'U', b'V', b'W',
-    b'X', b'Y', b'Z', b'[', b'\\', b']', b'^', b'_',
-    b'`',
-
-          b'A', b'B', b'C', b'D', b'E', b'F', b'G',
-    b'H', b'I', b'J', b'K', b'L', b'M', b'N', b'O',
-    b'P', b'Q', b'R', b'S', b'T', b'U', b'V', b'W',
-    b'X', b'Y', b'Z',
-
-                      b'{', b'|', b'}', b'~', 0x7f,
-    0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
-    0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,
-    0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
-    0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
-    0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
-    0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
-    0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
-    0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
-    0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
-    0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
-    0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
-    0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
-    0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
-    0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,
-    0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
-    0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
-];
-
 enum AsciiCharacterClass {
     C,  // control
     Cw, // control whitespace

From fbe34cc521a5d0f04dd6afadb5b200f0338f41d9 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 20:40:32 +0100
Subject: [PATCH 03/11] =?UTF-8?q?Add=20benchmark=20for=20not-quite-correct?=
 =?UTF-8?q?=20=E2=80=9Cfake=20SIMD=E2=80=9D=20make=5Fascii=5Fuppercase?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 src/libcore/benches/ascii_case.rs | 46 +++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/src/libcore/benches/ascii_case.rs b/src/libcore/benches/ascii_case.rs
index 1c9691d1ee173..df82a8cb1d537 100644
--- a/src/libcore/benches/ascii_case.rs
+++ b/src/libcore/benches/ascii_case.rs
@@ -87,6 +87,52 @@ benches! {
     fn bench06_libcore(bytes: &mut [u8]) {
         bytes.make_ascii_uppercase()
     }
+
+    fn bench07_fake_simd_u32(bytes: &mut [u8]) {
+        let (before, aligned, after) = unsafe {
+            bytes.align_to_mut::<u32>()
+        };
+        for byte in before {
+            *byte = branchless_to_ascii_upper_case(*byte)
+        }
+        for word in aligned {
+            // FIXME: this is incorrect for some byte values:
+            // addition within a byte can carry/overflow into the next byte.
+            // Test case: b"\xFFz  "
+            *word &= !(
+                (
+                    word.wrapping_add(0x1f1f1f1f) &
+                    !word.wrapping_add(0x05050505) &
+                    0x80808080
+                ) >> 2
+            )
+        }
+        for byte in after {
+            *byte = branchless_to_ascii_upper_case(*byte)
+        }
+    }
+
+    fn bench08_fake_simd_u64(bytes: &mut [u8]) {
+        let (before, aligned, after) = unsafe {
+            bytes.align_to_mut::<u64>()
+        };
+        for byte in before {
+            *byte = branchless_to_ascii_upper_case(*byte)
+        }
+        for word in aligned {
+            // FIXME: like above, this is incorrect for some byte values.
+            *word &= !(
+                (
+                    word.wrapping_add(0x1f1f1f1f_1f1f1f1f) &
+                    !word.wrapping_add(0x05050505_05050505) &
+                    0x80808080_80808080
+                ) >> 2
+            )
+        }
+        for byte in after {
+            *byte = branchless_to_ascii_upper_case(*byte)
+        }
+    }
 }
 
 macro_rules! repeat {

From e3fb6f89fef4fea94deeac572b0373843560c2d6 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 21:28:33 +0100
Subject: [PATCH 04/11] Tidy

---
 src/libcore/benches/ascii_case.rs | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/src/libcore/benches/ascii_case.rs b/src/libcore/benches/ascii_case.rs
index df82a8cb1d537..dcc87cc45b1e3 100644
--- a/src/libcore/benches/ascii_case.rs
+++ b/src/libcore/benches/ascii_case.rs
@@ -145,9 +145,11 @@ const LONG: &'static str = repeat!(r#"
     La Guida di Bragia, a Ballad Opera for the Marionette Theatre (around 1850)
     Alice's Adventures in Wonderland (1865)
     Phantasmagoria and Other Poems (1869)
-    Through the Looking-Glass, and What Alice Found There (includes "Jabberwocky" and "The Walrus and the Carpenter") (1871)
+    Through the Looking-Glass, and What Alice Found There
+        (includes "Jabberwocky" and "The Walrus and the Carpenter") (1871)
     The Hunting of the Snark (1876)
-    Rhyme? And Reason? (1883) – shares some contents with the 1869 collection, including the long poem "Phantasmagoria"
+    Rhyme? And Reason? (1883) – shares some contents with the 1869 collection,
+        including the long poem "Phantasmagoria"
     A Tangled Tale (1885)
     Sylvie and Bruno (1889)
     Sylvie and Bruno Concluded (1893)

From 525a043c6bc8a220b7dbdde3eb10bb79f8e99506 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 23:32:36 +0100
Subject: [PATCH 05/11] Rename src/libcore/benches/ascii_case.rs to ascii.rs

---
 src/libcore/benches/{ascii_case.rs => ascii.rs} | 6 ++++++
 src/libcore/benches/lib.rs                      | 2 +-
 2 files changed, 7 insertions(+), 1 deletion(-)
 rename src/libcore/benches/{ascii_case.rs => ascii.rs} (97%)

diff --git a/src/libcore/benches/ascii_case.rs b/src/libcore/benches/ascii.rs
similarity index 97%
rename from src/libcore/benches/ascii_case.rs
rename to src/libcore/benches/ascii.rs
index dcc87cc45b1e3..2714a8a5c278e 100644
--- a/src/libcore/benches/ascii_case.rs
+++ b/src/libcore/benches/ascii.rs
@@ -84,6 +84,12 @@ benches! {
         }
     }
 
+    fn bench05_multiply_by_bool(bytes: &mut [u8]) {
+        for byte in bytes {
+            *byte &= !(0x20 * (b'a' <= *byte && *byte <= b'z') as u8)
+        }
+    }
+
     fn bench06_libcore(bytes: &mut [u8]) {
         bytes.make_ascii_uppercase()
     }
diff --git a/src/libcore/benches/lib.rs b/src/libcore/benches/lib.rs
index b72d035b0ce94..707cdd5f450ea 100644
--- a/src/libcore/benches/lib.rs
+++ b/src/libcore/benches/lib.rs
@@ -5,7 +5,7 @@ extern crate core;
 extern crate test;
 
 mod any;
-mod ascii_case;
+mod ascii;
 mod char;
 mod hash;
 mod iter;

From 6d3840b23a185d057cc9792a2516e6a71887f277 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 23:56:50 +0100
Subject: [PATCH 06/11] Add benchmarks for `u8::is_ascii*`

---
 src/libcore/benches/ascii.rs | 49 +++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 17 deletions(-)

diff --git a/src/libcore/benches/ascii.rs b/src/libcore/benches/ascii.rs
index 2714a8a5c278e..e20b1b9bf3136 100644
--- a/src/libcore/benches/ascii.rs
+++ b/src/libcore/benches/ascii.rs
@@ -12,7 +12,14 @@ fn branchless_to_ascii_upper_case(byte: u8) -> u8 {
 
 
 macro_rules! benches {
-    ($( fn $name: ident($arg: ident: &mut [u8]) $body: block )+) => {
+    ($( fn $name: ident($arg: ident: &mut [u8]) $body: block )+ @iter $( $is_: ident, )+) => {
+        benches! {@
+            $( fn $name($arg: &mut [u8]) $body )+
+            $( fn $is_(bytes: &mut [u8]) { bytes.iter().all(u8::$is_) } )+
+        }
+    };
+
+    (@$( fn $name: ident($arg: ident: &mut [u8]) $body: block )+) => {
         benches!(mod short SHORT $($name $arg $body)+);
         benches!(mod medium MEDIUM $($name $arg $body)+);
         benches!(mod long LONG $($name $arg $body)+);
@@ -30,7 +37,7 @@ macro_rules! benches {
                         let mut vec = $input.as_bytes().to_vec();
                         {
                             let $arg = &mut vec[..];
-                            $body
+                            black_box($body);
                         }
                         vec
                     })
@@ -44,21 +51,21 @@ use test::black_box;
 use test::Bencher;
 
 benches! {
-    fn bench00_alloc_only(_bytes: &mut [u8]) {}
+    fn case00_alloc_only(_bytes: &mut [u8]) {}
 
-    fn bench01_black_box_read_each_byte(bytes: &mut [u8]) {
+    fn case01_black_box_read_each_byte(bytes: &mut [u8]) {
         for byte in bytes {
             black_box(*byte);
         }
     }
 
-    fn bench02_lookup(bytes: &mut [u8]) {
+    fn case02_lookup(bytes: &mut [u8]) {
         for byte in bytes {
             *byte = ASCII_UPPERCASE_MAP[*byte as usize]
         }
     }
 
-    fn bench03_branch_and_subtract(bytes: &mut [u8]) {
+    fn case03_branch_and_subtract(bytes: &mut [u8]) {
         for byte in bytes {
             *byte = if b'a' <= *byte && *byte <= b'z' {
                 *byte - b'a' + b'A'
@@ -68,7 +75,7 @@ benches! {
         }
     }
 
-    fn bench04_branch_and_mask(bytes: &mut [u8]) {
+    fn case04_branch_and_mask(bytes: &mut [u8]) {
         for byte in bytes {
             *byte = if b'a' <= *byte && *byte <= b'z' {
                 *byte & !0x20
@@ -78,23 +85,17 @@ benches! {
         }
     }
 
-    fn bench05_branchless(bytes: &mut [u8]) {
+    fn case05_branchless(bytes: &mut [u8]) {
         for byte in bytes {
             *byte = branchless_to_ascii_upper_case(*byte)
         }
     }
 
-    fn bench05_multiply_by_bool(bytes: &mut [u8]) {
-        for byte in bytes {
-            *byte &= !(0x20 * (b'a' <= *byte && *byte <= b'z') as u8)
-        }
-    }
-
-    fn bench06_libcore(bytes: &mut [u8]) {
+    fn case06_libcore(bytes: &mut [u8]) {
         bytes.make_ascii_uppercase()
     }
 
-    fn bench07_fake_simd_u32(bytes: &mut [u8]) {
+    fn case07_fake_simd_u32(bytes: &mut [u8]) {
         let (before, aligned, after) = unsafe {
             bytes.align_to_mut::<u32>()
         };
@@ -118,7 +119,7 @@ benches! {
         }
     }
 
-    fn bench08_fake_simd_u64(bytes: &mut [u8]) {
+    fn case08_fake_simd_u64(bytes: &mut [u8]) {
         let (before, aligned, after) = unsafe {
             bytes.align_to_mut::<u64>()
         };
@@ -139,6 +140,20 @@ benches! {
             *byte = branchless_to_ascii_upper_case(*byte)
         }
     }
+
+    @iter
+
+    is_ascii,
+    is_ascii_alphabetic,
+    is_ascii_uppercase,
+    is_ascii_lowercase,
+    is_ascii_alphanumeric,
+    is_ascii_digit,
+    is_ascii_hexdigit,
+    is_ascii_punctuation,
+    is_ascii_graphic,
+    is_ascii_whitespace,
+    is_ascii_control,
 }
 
 macro_rules! repeat {

From b4faa9b45635ae5185d562701d7429e5fbb08c8b Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Mon, 18 Mar 2019 23:57:09 +0100
Subject: [PATCH 07/11] Remove ASCII_CHARACTER_CLASS table, use `match` with
 range patterns instead.

---
 src/libcore/num/mod.rs | 75 +++++++++++-------------------------------
 1 file changed, 20 insertions(+), 55 deletions(-)

diff --git a/src/libcore/num/mod.rs b/src/libcore/num/mod.rs
index 977e21e0dc89a..64469a4b7e43c 100644
--- a/src/libcore/num/mod.rs
+++ b/src/libcore/num/mod.rs
@@ -3958,9 +3958,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_alphabetic(&self) -> bool {
-        if *self >= 0x80 { return false; }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            L | Lx | U | Ux => true,
+        match *self {
+            b'A'...b'Z' | b'a'...b'z' => true,
             _ => false
         }
     }
@@ -3994,9 +3993,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_uppercase(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            U | Ux => true,
+        match *self {
+            b'A'...b'Z' => true,
             _ => false
         }
     }
@@ -4030,9 +4028,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_lowercase(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            L | Lx => true,
+        match *self {
+            b'a'...b'z' => true,
             _ => false
         }
     }
@@ -4069,9 +4066,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_alphanumeric(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            D | L | Lx | U | Ux => true,
+        match *self {
+            b'0'...b'9' | b'A'...b'Z' | b'a'...b'z' => true,
             _ => false
         }
     }
@@ -4105,9 +4101,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_digit(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            D => true,
+        match *self {
+            b'0'...b'9' => true,
             _ => false
         }
     }
@@ -4144,9 +4139,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_hexdigit(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            D | Lx | Ux => true,
+        match *self {
+            b'0'...b'9' | b'A'...b'F' | b'a'...b'f' => true,
             _ => false
         }
     }
@@ -4184,9 +4178,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_punctuation(&self) -> bool {
-        if *self >= 0x80 { return false }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            P => true,
+        match *self {
+            b'!'...b'/' | b':'...b'@' | b'['...b'`' | b'{'...b'~' => true,
             _ => false
         }
     }
@@ -4220,9 +4213,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_graphic(&self) -> bool {
-        if *self >= 0x80 { return false; }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            Ux | U | Lx | L | D | P => true,
+        match *self {
+            b'!'...b'~' => true,
             _ => false
         }
     }
@@ -4273,9 +4265,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_whitespace(&self) -> bool {
-        if *self >= 0x80 { return false; }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            Cw | W => true,
+        match *self {
+            b'\t' | b'\n' | b'\x0C' | b'\r' | b' ' => true,
             _ => false
         }
     }
@@ -4311,9 +4302,8 @@ impl u8 {
     #[stable(feature = "ascii_ctype_on_intrinsics", since = "1.24.0")]
     #[inline]
     pub fn is_ascii_control(&self) -> bool {
-        if *self >= 0x80 { return false; }
-        match ASCII_CHARACTER_CLASS[*self as usize] {
-            C | Cw => true,
+        match *self {
+            b'\0'...b'\x1F' | b'\x7F' => true,
             _ => false
         }
     }
@@ -4979,28 +4969,3 @@ impl_from! { u32, f64, #[stable(feature = "lossless_float_conv", since = "1.6.0"
 
 // Float -> Float
 impl_from! { f32, f64, #[stable(feature = "lossless_float_conv", since = "1.6.0")] }
-
-enum AsciiCharacterClass {
-    C,  // control
-    Cw, // control whitespace
-    W,  // whitespace
-    D,  // digit
-    L,  // lowercase
-    Lx, // lowercase hex digit
-    U,  // uppercase
-    Ux, // uppercase hex digit
-    P,  // punctuation
-}
-use self::AsciiCharacterClass::*;
-
-static ASCII_CHARACTER_CLASS: [AsciiCharacterClass; 128] = [
-//  _0 _1 _2 _3 _4 _5 _6 _7 _8 _9 _a _b _c _d _e _f
-    C, C, C, C, C, C, C, C, C, Cw,Cw,C, Cw,Cw,C, C, // 0_
-    C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, // 1_
-    W, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, // 2_
-    D, D, D, D, D, D, D, D, D, D, P, P, P, P, P, P, // 3_
-    P, Ux,Ux,Ux,Ux,Ux,Ux,U, U, U, U, U, U, U, U, U, // 4_
-    U, U, U, U, U, U, U, U, U, U, U, P, P, P, P, P, // 5_
-    P, Lx,Lx,Lx,Lx,Lx,Lx,L, L, L, L, L, L, L, L, L, // 6_
-    L, L, L, L, L, L, L, L, L, L, L, P, P, P, P, C, // 7_
-];

From 4a3241a815e951dfa87e9be7133e8265bc945e57 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Tue, 19 Mar 2019 00:34:27 +0100
Subject: [PATCH 08/11] Benchmark more possibles impls of
 [u8]::make_ascii_uppercase

---
 src/libcore/benches/ascii.rs | 84 +++++++++++++++++++++++++++++++++++-
 1 file changed, 83 insertions(+), 1 deletion(-)

diff --git a/src/libcore/benches/ascii.rs b/src/libcore/benches/ascii.rs
index e20b1b9bf3136..ce36027394a48 100644
--- a/src/libcore/benches/ascii.rs
+++ b/src/libcore/benches/ascii.rs
@@ -59,7 +59,7 @@ benches! {
         }
     }
 
-    fn case02_lookup(bytes: &mut [u8]) {
+    fn case02_lookup_table(bytes: &mut [u8]) {
         for byte in bytes {
             *byte = ASCII_UPPERCASE_MAP[*byte as usize]
         }
@@ -141,6 +141,55 @@ benches! {
         }
     }
 
+    fn case09_mask_mult_bool_branchy_lookup_table(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            if b >= 0x80 { return false }
+            match ASCII_CHARACTER_CLASS[b as usize] {
+                L | Lx => true,
+                _ => false,
+            }
+        }
+        for byte in bytes {
+            *byte &= !(0x20 * (is_ascii_lowercase(*byte) as u8))
+        }
+    }
+
+    fn case10_mask_mult_bool_lookup_table(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            match ASCII_CHARACTER_CLASS[b as usize] {
+                L | Lx => true,
+                _ => false
+            }
+        }
+        for byte in bytes {
+            *byte &= !(0x20 * (is_ascii_lowercase(*byte) as u8))
+        }
+    }
+
+    fn case11_mask_mult_bool_match_range(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            match b {
+                b'a'...b'z' => true,
+                _ => false
+            }
+        }
+        for byte in bytes {
+            *byte &= !(0x20 * (is_ascii_lowercase(*byte) as u8))
+        }
+    }
+
+    fn case12_mask_shifted_bool_match_range(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            match b {
+                b'a'...b'z' => true,
+                _ => false
+            }
+        }
+        for byte in bytes {
+            *byte &= !((is_ascii_lowercase(*byte) as u8) << 5)
+        }
+    }
+
     @iter
 
     is_ascii,
@@ -219,3 +268,36 @@ const ASCII_UPPERCASE_MAP: [u8; 256] = [
     0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
 ];
 
+enum AsciiCharacterClass {
+    C,  // control
+    Cw, // control whitespace
+    W,  // whitespace
+    D,  // digit
+    L,  // lowercase
+    Lx, // lowercase hex digit
+    U,  // uppercase
+    Ux, // uppercase hex digit
+    P,  // punctuation
+    N,  // Non-ASCII
+}
+use self::AsciiCharacterClass::*;
+
+static ASCII_CHARACTER_CLASS: [AsciiCharacterClass; 256] = [
+//  _0 _1 _2 _3 _4 _5 _6 _7 _8 _9 _a _b _c _d _e _f
+    C, C, C, C, C, C, C, C, C, Cw,Cw,C, Cw,Cw,C, C, // 0_
+    C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, // 1_
+    W, P, P, P, P, P, P, P, P, P, P, P, P, P, P, P, // 2_
+    D, D, D, D, D, D, D, D, D, D, P, P, P, P, P, P, // 3_
+    P, Ux,Ux,Ux,Ux,Ux,Ux,U, U, U, U, U, U, U, U, U, // 4_
+    U, U, U, U, U, U, U, U, U, U, U, P, P, P, P, P, // 5_
+    P, Lx,Lx,Lx,Lx,Lx,Lx,L, L, L, L, L, L, L, L, L, // 6_
+    L, L, L, L, L, L, L, L, L, L, L, P, P, P, P, C, // 7_
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+    N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+];

From 0ad91f73d92c3b8d3978f8f54c04b8efe3d2e673 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Tue, 19 Mar 2019 00:50:26 +0100
Subject: [PATCH 09/11] Simplify u8::to_ascii_{upp,low}ercase while keeping it
 fast

---
 src/libcore/benches/ascii.rs | 24 ++++++++++++++++++-
 src/libcore/num/mod.rs       | 46 ++++--------------------------------
 2 files changed, 27 insertions(+), 43 deletions(-)

diff --git a/src/libcore/benches/ascii.rs b/src/libcore/benches/ascii.rs
index ce36027394a48..89e67cca4b754 100644
--- a/src/libcore/benches/ascii.rs
+++ b/src/libcore/benches/ascii.rs
@@ -1,4 +1,26 @@
-// See comments in `u8::to_ascii_uppercase` in `src/libcore/num/mod.rs`.
+// Lower-case ASCII 'a' is the first byte that has its highest bit set
+// after wrap-adding 0x1F:
+//
+//     b'a' + 0x1F == 0x80 == 0b1000_0000
+//     b'z' + 0x1F == 0x98 == 0b10011000
+//
+// Lower-case ASCII 'z' is the last byte that has its highest bit unset
+// after wrap-adding 0x05:
+//
+//     b'a' + 0x05 == 0x66 == 0b0110_0110
+//     b'z' + 0x05 == 0x7F == 0b0111_1111
+//
+// … except for 0xFB to 0xFF, but those are in the range of bytes
+// that have the highest bit unset again after adding 0x1F.
+//
+// So `(byte + 0x1f) & !(byte + 5)` has its highest bit set
+// iff `byte` is a lower-case ASCII letter.
+//
+// Lower-case ASCII letters all have the 0x20 bit set.
+// (Two positions right of 0x80, the highest bit.)
+// Unsetting that bit produces the same letter, in upper-case.
+//
+// Therefore:
 fn branchless_to_ascii_upper_case(byte: u8) -> u8 {
     byte &
     !(
diff --git a/src/libcore/num/mod.rs b/src/libcore/num/mod.rs
index 64469a4b7e43c..3fcae6b94b06d 100644
--- a/src/libcore/num/mod.rs
+++ b/src/libcore/num/mod.rs
@@ -3794,39 +3794,8 @@ impl u8 {
     #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
     #[inline]
     pub fn to_ascii_uppercase(&self) -> u8 {
-        // See benchmarks in src/libcore/benches/ascii_case.rs
-
-        // Lower-case ASCII 'a' is the first byte that has its highest bit set
-        // after wrap-adding 0x1F:
-        //
-        //     b'a' + 0x1F == 0x80 == 0b1000_0000
-        //     b'z' + 0x1F == 0x98 == 0b10011000
-        //
-        // Lower-case ASCII 'z' is the last byte that has its highest bit unset
-        // after wrap-adding 0x05:
-        //
-        //     b'a' + 0x05 == 0x66 == 0b0110_0110
-        //     b'z' + 0x05 == 0x7F == 0b0111_1111
-        //
-        // … except for 0xFB to 0xFF, but those are in the range of bytes
-        // that have the highest bit unset again after adding 0x1F.
-        //
-        // So `(byte + 0x1f) & !(byte + 5)` has its highest bit set
-        // iff `byte` is a lower-case ASCII letter.
-        //
-        // Lower-case ASCII letters all have the 0x20 bit set.
-        // (Two positions right of 0x80, the highest bit.)
-        // Unsetting that bit produces the same letter, in upper-case.
-        //
-        // Therefore:
-        *self &
-        !(
-            (
-                self.wrapping_add(0x1f) &
-                !self.wrapping_add(0x05) &
-                0x80
-            ) >> 2
-        )
+        // Unset the fith bit if this is a lowercase letter
+        *self & !((self.is_ascii_lowercase() as u8) << 5)
     }
 
     /// Makes a copy of the value in its ASCII lower case equivalent.
@@ -3848,15 +3817,8 @@ impl u8 {
     #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
     #[inline]
     pub fn to_ascii_lowercase(&self) -> u8 {
-        // See comments in to_ascii_uppercase above.
-        *self |
-        (
-            (
-                self.wrapping_add(0x3f) &
-                !self.wrapping_add(0x25) &
-                0x80
-            ) >> 2
-        )
+        // Set the fith bit if this is an uppercase letter
+        *self | ((self.is_ascii_uppercase() as u8) << 5)
     }
 
     /// Checks that two values are an ASCII case-insensitive match.

From c1ec29ace000ce4c733cde6948c87f9bc2370691 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Tue, 19 Mar 2019 08:32:15 +0100
Subject: [PATCH 10/11] ASCII uppercase: add "subtract shifted bool" benchmark

---
 src/libcore/benches/ascii.rs | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/src/libcore/benches/ascii.rs b/src/libcore/benches/ascii.rs
index 89e67cca4b754..968d28a995897 100644
--- a/src/libcore/benches/ascii.rs
+++ b/src/libcore/benches/ascii.rs
@@ -212,6 +212,18 @@ benches! {
         }
     }
 
+    fn case13_subtract_shifted_bool_match_range(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            match b {
+                b'a'...b'z' => true,
+                _ => false
+            }
+        }
+        for byte in bytes {
+            *byte -= (is_ascii_lowercase(*byte) as u8) << 5
+        }
+    }
+
     @iter
 
     is_ascii,

From 7fad370fe9dc000f6f6f497a1939de6196e2c4fc Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon.sapin@exyr.org>
Date: Tue, 19 Mar 2019 13:41:59 +0100
Subject: [PATCH 11/11] ASCII uppercase: add "subtract multiplied bool"
 benchmark

---
 src/libcore/benches/ascii.rs | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/src/libcore/benches/ascii.rs b/src/libcore/benches/ascii.rs
index 968d28a995897..635537e312158 100644
--- a/src/libcore/benches/ascii.rs
+++ b/src/libcore/benches/ascii.rs
@@ -224,6 +224,18 @@ benches! {
         }
     }
 
+    fn case14_subtract_multiplied_bool_match_range(bytes: &mut [u8]) {
+        fn is_ascii_lowercase(b: u8) -> bool {
+            match b {
+                b'a'...b'z' => true,
+                _ => false
+            }
+        }
+        for byte in bytes {
+            *byte -= (b'a' - b'A') * is_ascii_lowercase(*byte) as u8
+        }
+    }
+
     @iter
 
     is_ascii,