Align EMOJI_PRESENTATION_LEAVES to 128 bytes

Ensure rows don't cross cache lines, makes a small difference in the benchmarks
unicode-rs · Mar 5, 2024 · 4aa5fb8 · 4aa5fb8
1 parent 75be2e9
commit 4aa5fb8
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 11 deletions.
diff --git a/scripts/unicode.py b/scripts/unicode.py
@@ -488,8 +488,6 @@ def emit_module(
         module.write(
             """
 pub mod charwidth {
-    use core::option::Option::{self, None, Some};
-
     /// Returns the [UAX #11](https://www.unicode.org/reports/tr11/) based width of `c` by
     /// consulting a multi-level lookup table.
     /// If `is_cjk == true`, ambiguous width characters are treated as double width; otherwise,
@@ -563,7 +561,7 @@ def emit_module(
         // Extract the 3-9th (0-indexed) least significant bits of `cp`,
         // and use them to index into `leaf_row`.
         let idx_within_leaf = usize::try_from((cp >> 3) & 0x7F).unwrap();
-        let leaf_byte = EMOJI_PRESENTATION_LEAVES[idx_of_leaf][idx_within_leaf];
+        let leaf_byte = EMOJI_PRESENTATION_LEAVES.0[idx_of_leaf][idx_within_leaf];
 
         // Use the 3 LSB of `cp` to index into `leaf_byte`.
         ((leaf_byte >> (cp & 7)) & 1) == 1
@@ -624,9 +622,12 @@ def emit_module(
 
         module.write(
             f"""
+    #[repr(align(128))]
+    struct Align128<T>(T);
+
     /// Array of 1024-bit bitmaps. Index into the correct (obtained from `EMOJI_PRESENTATION_INDEX`)
     /// bitmap with the 10 LSB of your codepoint to get whether it can start an emoji presentation seq.
-    static EMOJI_PRESENTATION_LEAVES: [[u8; 128]; {len(variation_leaves)}] = [
+    static EMOJI_PRESENTATION_LEAVES: Align128<[[u8; 128]; {len(variation_leaves)}]> = Align128([
 """
         )
         for leaf in variation_leaves:
@@ -638,7 +639,7 @@ def emit_module(
                 module.write("\n")
             module.write("        ],\n")
 
-        module.write("    ];\n")
+        module.write("    ]);\n")
         module.write("}\n")
 
 

diff --git a/src/tables.rs b/src/tables.rs
@@ -15,8 +15,6 @@
 pub const UNICODE_VERSION: (u8, u8, u8) = (15, 1, 0);
 
 pub mod charwidth {
-    use core::option::Option::{self, None, Some};
-
     /// Returns the [UAX #11](https://www.unicode.org/reports/tr11/) based width of `c` by
     /// consulting a multi-level lookup table.
     /// If `is_cjk == true`, ambiguous width characters are treated as double width; otherwise,
@@ -83,7 +81,7 @@ pub mod charwidth {
         // Extract the 3-9th (0-indexed) least significant bits of `cp`,
         // and use them to index into `leaf_row`.
         let idx_within_leaf = usize::try_from((cp >> 3) & 0x7F).unwrap();
-        let leaf_byte = EMOJI_PRESENTATION_LEAVES[idx_of_leaf][idx_within_leaf];
+        let leaf_byte = EMOJI_PRESENTATION_LEAVES.0[idx_of_leaf][idx_within_leaf];
 
         // Use the 3 LSB of `cp` to index into `leaf_byte`.
         ((leaf_byte >> (cp & 7)) & 1) == 1
@@ -568,9 +566,12 @@ pub mod charwidth {
         0xFF, 0xFF, 0x5F,
     ];
 
+    #[repr(align(128))]
+    struct Align128<T>(T);
+
     /// Array of 1024-bit bitmaps. Index into the correct (obtained from `EMOJI_PRESENTATION_INDEX`)
     /// bitmap with the 10 LSB of your codepoint to get whether it can start an emoji presentation seq.
-    static EMOJI_PRESENTATION_LEAVES: [[u8; 128]; 6] = [
+    static EMOJI_PRESENTATION_LEAVES: Align128<[[u8; 128]; 6]> = Align128([
         [
             0x00, 0x00, 0x00, 0x00, 0x08, 0x04, 0xFF, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
             0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x42, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
@@ -643,5 +644,5 @@ pub mod charwidth {
             0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xFF, 0x0F,
             0x01, 0x00,
         ],
-    ];
+    ]);
 }
diff --git a/src/tests.rs b/src/tests.rs
@@ -15,7 +15,7 @@ use std::iter;
 use std::prelude::v1::*;
 
 #[cfg(feature = "bench")]
-use test::{self, Bencher};
+use test::Bencher;
 
 #[cfg(feature = "bench")]
 #[bench]