From 1fe539f1d443336af26094086ea72083a15e92cc Mon Sep 17 00:00:00 2001
From: Nicolas Boichat <nicolas@boichat.ch>
Date: Wed, 14 May 2025 14:12:35 +0800
Subject: [PATCH 1/2] wc: Increase buffer size to 256kb

Improves performance by about 4% on large files.
---
 src/uu/wc/src/count_fast.rs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/uu/wc/src/count_fast.rs b/src/uu/wc/src/count_fast.rs
index b79e6b0e3..450372d4d 100644
--- a/src/uu/wc/src/count_fast.rs
+++ b/src/uu/wc/src/count_fast.rs
@@ -32,7 +32,7 @@ use libc::S_IFIFO;
 #[cfg(any(target_os = "linux", target_os = "android"))]
 use uucore::pipes::{pipe, splice, splice_exact};
 
-const BUF_SIZE: usize = 16 * 1024;
+const BUF_SIZE: usize = 256 * 1024;
 #[cfg(any(target_os = "linux", target_os = "android"))]
 const SPLICE_SIZE: usize = 128 * 1024;
 

From 1fc14d837c54ca3d163efa60d83753628fbf8bc7 Mon Sep 17 00:00:00 2001
From: Nicolas Boichat <nicolas@boichat.ch>
Date: Wed, 14 May 2025 20:24:06 +0800
Subject: [PATCH 2/2] wc: Align buffer to 32-byte boundary

bytecount uses vector operations to speed up line counting.
At least on x86 with AVX2 support, the vectors are 256-byte wide,
and operations are much faster if the data is aligned.

Saves about 4% of total performance, matching wc's performance.
---
 src/uu/wc/src/count_fast.rs | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/src/uu/wc/src/count_fast.rs b/src/uu/wc/src/count_fast.rs
index 450372d4d..479183263 100644
--- a/src/uu/wc/src/count_fast.rs
+++ b/src/uu/wc/src/count_fast.rs
@@ -197,6 +197,23 @@ pub(crate) fn count_bytes_fast<T: WordCountable>(handle: &mut T) -> (usize, Opti
     }
 }
 
+/// A simple structure used to align a BUF_SIZE buffer to 32-byte boundary.
+///
+/// This is useful as bytecount uses 256-bit wide vector operations that run much
+/// faster on aligned data (at least on x86 with AVX2 support).
+#[repr(align(32))]
+struct AlignedBuffer {
+    data: [u8; BUF_SIZE],
+}
+
+impl Default for AlignedBuffer {
+    fn default() -> Self {
+        Self {
+            data: [0; BUF_SIZE],
+        }
+    }
+}
+
 /// Returns a WordCount that counts the number of bytes, lines, and/or the number of Unicode characters encoded in UTF-8 read via a Reader.
 ///
 /// This corresponds to the `-c`, `-l` and `-m` command line flags to wc.
@@ -213,9 +230,9 @@ pub(crate) fn count_bytes_chars_and_lines_fast<
     handle: &mut R,
 ) -> (WordCount, Option<io::Error>) {
     let mut total = WordCount::default();
-    let mut buf = [0; BUF_SIZE];
+    let buf: &mut [u8] = &mut AlignedBuffer::default().data;
     loop {
-        match handle.read(&mut buf) {
+        match handle.read(buf) {
             Ok(0) => return (total, None),
             Ok(n) => {
                 if COUNT_BYTES {