Add hash benchmarks

benches/all_benchmark.rs

@@ -3,6 +3,7 @@ use criterion::criterion_main;
 mod benchmarks;
 
 criterion_main! {
-    benchmarks::fasta_benchmark::fasta_benches,
     benchmarks::dna_benchmark::dna_benches,
+    benchmarks::fasta_benchmark::fasta_benches,
+    benchmarks::hash_benchmark::hash_benches,
 }

benches/benchmarks/hash_benchmark.rs

@@ -0,0 +1,18 @@
+use criterion::{criterion_group, criterion_main, Criterion};
+use nuc::dna::Dna;
+
+pub fn hash_nucleotides_benchmark(c: &mut Criterion) {
+    let dna = Dna::random(1_000_000);
+    let bytes = dna.as_bytes();
+    c.bench_function("Hashing random nucleotides", |b| {
+        b.iter(|| {
+            bytes
+                .chunks(4)
+                .map(|chunk| nuc::hash::hash_chars_be(chunk))
+                .count();
+        })
+    });
+}
+
+criterion_group!(hash_benches, hash_nucleotides_benchmark);
+criterion_main!(hash_benches);

benches/benchmarks/mod.rs

@@ -1,2 +1,3 @@
 pub mod fasta_benchmark;
 pub mod dna_benchmark;
+pub mod hash_benchmark;

src/hash.rs

@@ -1,6 +1,4 @@
-use std::sync::LazyLock;
-
-pub static CHAR_TO_TWO_BIT: LazyLock<[u8; 256]> = LazyLock::new(|| {
+pub const CHAR_TO_TWO_BIT: [u8; 256] = {
     let mut cache = [0; 256];
     cache[99] = 1; // lowercase c
     cache[67] = 1; // uppercase C
@@ -9,15 +7,24 @@ pub static CHAR_TO_TWO_BIT: LazyLock<[u8; 256]> = LazyLock::new(|| {
     cache[116] = 3; // lowercase t
     cache[84] = 3; // uppercase T
     cache
-});
+};
 
 /// Hashes a byte slice into a 8-bit integer.
 pub fn hash_chars_be(bytes: &[u8]) -> u8 {
     match bytes.len() {
         1 => CHAR_TO_TWO_BIT[bytes[0] as usize] << 6,
         2 => CHAR_TO_TWO_BIT[bytes[0] as usize] << 6 | (CHAR_TO_TWO_BIT[bytes[1] as usize] << 4),
-        3 => CHAR_TO_TWO_BIT[bytes[0] as usize] << 6 | (CHAR_TO_TWO_BIT[bytes[1] as usize] << 4) | (CHAR_TO_TWO_BIT[bytes[2] as usize] << 2),
-        _ => CHAR_TO_TWO_BIT[bytes[0] as usize] << 6 | (CHAR_TO_TWO_BIT[bytes[1] as usize] << 4) | (CHAR_TO_TWO_BIT[bytes[2] as usize] << 2) | (CHAR_TO_TWO_BIT[bytes[3] as usize]),
+        3 => {
+            CHAR_TO_TWO_BIT[bytes[0] as usize] << 6
+                | (CHAR_TO_TWO_BIT[bytes[1] as usize] << 4)
+                | (CHAR_TO_TWO_BIT[bytes[2] as usize] << 2)
+        }
+        _ => {
+            CHAR_TO_TWO_BIT[bytes[0] as usize] << 6
+                | (CHAR_TO_TWO_BIT[bytes[1] as usize] << 4)
+                | (CHAR_TO_TWO_BIT[bytes[2] as usize] << 2)
+                | (CHAR_TO_TWO_BIT[bytes[3] as usize])
+        }
     }
 }