From c9139d92ffa7a3c94c31b7df8e40c148b823f090 Mon Sep 17 00:00:00 2001
From: behzad nouri <behzadnouri@gmail.com>
Date: Thu, 19 Dec 2024 09:12:02 -0600
Subject: [PATCH 1/2] builds deduper hashers once per reset cycle

The commit avoids RandomState::build_hasher on each item and instead
just clones initialized hahsers which are pretty cheap to clone:
https://github.com/tkaitchuck/aHash/blob/7d5c661a7/src/aes_hash.rs#L18-L23
---
 perf/src/deduper.rs | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/perf/src/deduper.rs b/perf/src/deduper.rs
index b25bcc93ed4449..2a985b9c7a0383 100644
--- a/perf/src/deduper.rs
+++ b/perf/src/deduper.rs
@@ -2,7 +2,7 @@
 
 use {
     crate::packet::PacketBatch,
-    ahash::RandomState,
+    ahash::{AHasher, RandomState},
     rand::Rng,
     std::{
         hash::{BuildHasher, Hash, Hasher},
@@ -16,7 +16,7 @@ use {
 pub struct Deduper<const K: usize, T: ?Sized> {
     num_bits: u64,
     bits: Vec<AtomicU64>,
-    state: [RandomState; K],
+    hashers: [AHasher; K],
     clock: Instant,
     popcount: AtomicU64, // Number of one bits in self.bits.
     _phantom: PhantomData<T>,
@@ -28,7 +28,7 @@ impl<const K: usize, T: ?Sized + Hash> Deduper<K, T> {
         let size = usize::try_from(size).unwrap();
         Self {
             num_bits,
-            state: std::array::from_fn(|_| new_random_state(rng)),
+            hashers: std::array::from_fn(|_| new_hasher(rng)),
             clock: Instant::now(),
             bits: repeat_with(AtomicU64::default).take(size).collect(),
             popcount: AtomicU64::default(),
@@ -54,7 +54,7 @@ impl<const K: usize, T: ?Sized + Hash> Deduper<K, T> {
         assert!(0.0 < false_positive_rate && false_positive_rate < 1.0);
         let saturated = self.false_positive_rate() >= false_positive_rate;
         if saturated || self.clock.elapsed() >= reset_cycle {
-            self.state = std::array::from_fn(|_| new_random_state(rng));
+            self.hashers = std::array::from_fn(|_| new_hasher(rng));
             self.clock = Instant::now();
             self.bits.fill_with(AtomicU64::default);
             self.popcount = AtomicU64::default();
@@ -67,8 +67,7 @@ impl<const K: usize, T: ?Sized + Hash> Deduper<K, T> {
     #[allow(clippy::arithmetic_side_effects)]
     pub fn dedup(&self, data: &T) -> bool {
         let mut out = true;
-        let hashers = self.state.iter().map(RandomState::build_hasher);
-        for mut hasher in hashers {
+        for mut hasher in self.hashers.iter().map(AHasher::clone) {
             data.hash(&mut hasher);
             let hash: u64 = hasher.finish() % self.num_bits;
             let index = (hash >> 6) as usize;
@@ -83,8 +82,8 @@ impl<const K: usize, T: ?Sized + Hash> Deduper<K, T> {
     }
 }
 
-fn new_random_state<R: Rng>(rng: &mut R) -> RandomState {
-    RandomState::with_seeds(rng.gen(), rng.gen(), rng.gen(), rng.gen())
+fn new_hasher<R: Rng>(rng: &mut R) -> AHasher {
+    RandomState::with_seeds(rng.gen(), rng.gen(), rng.gen(), rng.gen()).build_hasher()
 }
 
 pub fn dedup_packets_and_count_discards<const K: usize>(

From 1a042860d244a400be9bfe625de9edf2fcfb72ba Mon Sep 17 00:00:00 2001
From: behzad nouri <behzadnouri@gmail.com>
Date: Thu, 19 Dec 2024 11:56:37 -0600
Subject: [PATCH 2/2] clones inside the loop

---
 perf/src/deduper.rs | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/perf/src/deduper.rs b/perf/src/deduper.rs
index 2a985b9c7a0383..36213cf1f65877 100644
--- a/perf/src/deduper.rs
+++ b/perf/src/deduper.rs
@@ -67,7 +67,8 @@ impl<const K: usize, T: ?Sized + Hash> Deduper<K, T> {
     #[allow(clippy::arithmetic_side_effects)]
     pub fn dedup(&self, data: &T) -> bool {
         let mut out = true;
-        for mut hasher in self.hashers.iter().map(AHasher::clone) {
+        for hasher in &self.hashers {
+            let mut hasher = hasher.clone();
             data.hash(&mut hasher);
             let hash: u64 = hasher.finish() % self.num_bits;
             let index = (hash >> 6) as usize;