perf: Miscellaneous router perf improvements (#7477)

Signed-off-by: jthomson04 <jwillthomson19@gmail.com>

lib/kv-router/src/protocols.rs

@@ -31,6 +31,29 @@ pub struct BlockHashOptions<'a> {
     pub is_eagle: Option<bool>,
 }
 
+#[inline]
+fn hash_block_no_mm(chunk: &[u32], seed: u64, scratch_bytes: &mut Vec<u8>) -> LocalBlockHash {
+    #[cfg(target_endian = "little")]
+    {
+        let _ = scratch_bytes;
+        // SAFETY: `u32` is plain-old-data, and on little-endian targets its in-memory
+        // representation matches the `to_le_bytes()` sequence used for hashing.
+        let chunk_bytes = unsafe {
+            std::slice::from_raw_parts(chunk.as_ptr().cast::<u8>(), std::mem::size_of_val(chunk))
+        };
+        LocalBlockHash(xxh3::xxh3_64_with_seed(chunk_bytes, seed))
+    }
+
+    #[cfg(not(target_endian = "little"))]
+    {
+        scratch_bytes.clear();
+        for &token in chunk {
+            scratch_bytes.extend_from_slice(&token.to_le_bytes());
+        }
+        LocalBlockHash(xxh3::xxh3_64_with_seed(scratch_bytes, seed))
+    }
+}
+
 /// Compute the hash for a sequence of tokens, optionally including multimodal metadata
 /// and LoRA adapter identity.
 ///
@@ -56,35 +79,42 @@ pub fn compute_block_hash_for_seq(
         Some(name) => XXH3_SEED.wrapping_add(xxh3::xxh3_64(name.as_bytes())),
         None => XXH3_SEED,
     };
-
     let is_eagle_flag = options.is_eagle.unwrap_or(false);
     let stride = kv_block_size as usize;
     let window_size = if is_eagle_flag { stride + 1 } else { stride };
-
-    let mut hashes = Vec::new();
+    let estimated_blocks = if is_eagle_flag {
+        tokens.len().saturating_sub(1) / stride
+    } else {
+        tokens.len() / stride
+    };
+    let mut hashes = Vec::with_capacity(estimated_blocks);
+    let mut bytes = Vec::with_capacity(window_size * std::mem::size_of::<u32>());
+    let mut mm_hashes = Vec::new();
     let mut block_idx = 0;
     let mut start = 0;
 
     while start + window_size <= tokens.len() {
         let chunk = &tokens[start..start + window_size];
-        let mut bytes: Vec<u8> = chunk.iter().flat_map(|&num| num.to_le_bytes()).collect();
-
         if let Some(mm_infos) = options.block_mm_infos
             && let Some(Some(block_mm_info)) = mm_infos.get(block_idx)
         {
-            let mut mm_hashes: Vec<u64> = block_mm_info
-                .mm_objects
-                .iter()
-                .map(|obj| obj.mm_hash)
-                .collect();
+            bytes.clear();
+            for &token in chunk {
+                bytes.extend_from_slice(&token.to_le_bytes());
+            }
+
+            mm_hashes.clear();
+            mm_hashes.extend(block_mm_info.mm_objects.iter().map(|obj| obj.mm_hash));
             mm_hashes.sort_unstable();
 
-            for mm_hash in mm_hashes {
+            for &mm_hash in &mm_hashes {
                 bytes.extend_from_slice(&mm_hash.to_le_bytes());
             }
-        }
 
-        hashes.push(LocalBlockHash(xxh3::xxh3_64_with_seed(&bytes, seed)));
+            hashes.push(LocalBlockHash(xxh3::xxh3_64_with_seed(&bytes, seed)));
+        } else {
+            hashes.push(hash_block_no_mm(chunk, seed, &mut bytes));
+        }
 
         start += stride;
         block_idx += 1;
@@ -110,8 +140,25 @@ pub fn compute_seq_hash_for_block(block_hashes: &[LocalBlockHash]) -> Vec<Sequen
         let current_block_hash = block_hashes[i].0;
 
         let combined = [parent_seq_hash, current_block_hash];
-        let bytes: Vec<u8> = combined.iter().flat_map(|&num| num.to_le_bytes()).collect();
-        let seq_hash = compute_hash(&bytes);
+        #[cfg(target_endian = "little")]
+        let seq_hash = {
+            // SAFETY: `u64` is plain-old-data, and on little-endian targets its in-memory
+            // representation matches the `to_le_bytes()` sequence used by the previous code.
+            let bytes = unsafe {
+                std::slice::from_raw_parts(
+                    combined.as_ptr().cast::<u8>(),
+                    std::mem::size_of_val(&combined),
+                )
+            };
+            compute_hash(bytes)
+        };
+        #[cfg(not(target_endian = "little"))]
+        let seq_hash = {
+            let mut bytes = [0_u8; std::mem::size_of::<u64>() * 2];
+            bytes[..8].copy_from_slice(&parent_seq_hash.to_le_bytes());
+            bytes[8..].copy_from_slice(&current_block_hash.to_le_bytes());
+            compute_hash(&bytes)
+        };
         sequence_hashes.push(seq_hash);
     }
 

lib/kv-router/src/scheduling/config.rs

@@ -9,7 +9,9 @@ use rand::Rng;
 use serde::{Deserialize, Serialize};
 use validator::{Validate, ValidationError};
 
-use crate::protocols::{BlockHashOptions, compute_block_hash_for_seq, compute_seq_hash_for_block};
+use crate::protocols::{
+    BlockHashOptions, LocalBlockHash, compute_block_hash_for_seq, compute_seq_hash_for_block,
+};
 
 const fn default_min_initial_workers() -> usize {
     1
@@ -218,6 +220,7 @@ impl KvRouterConfig {
         block_size: u32,
         config_override: Option<&RouterConfigOverride>,
         hash_options: BlockHashOptions<'_>,
+        precomputed_block_hashes: Option<&[LocalBlockHash]>,
     ) -> Option<Vec<u64>> {
         if !self.router_track_active_blocks {
             return None;
@@ -233,8 +236,14 @@ impl KvRouterConfig {
             .unwrap_or(self.router_assume_kv_reuse);
 
         if assume_kv_reuse {
-            let block_hashes = compute_block_hash_for_seq(tokens, block_size, hash_options);
-            Some(compute_seq_hash_for_block(&block_hashes))
+            let block_hashes = match precomputed_block_hashes {
+                Some(block_hashes) => block_hashes,
+                None => {
+                    let computed = compute_block_hash_for_seq(tokens, block_size, hash_options);
+                    return Some(compute_seq_hash_for_block(&computed));
+                }
+            };
+            Some(compute_seq_hash_for_block(block_hashes))
         } else {
             let mut rng = rand::rng();
             Some((0..num_blocks).map(|_| rng.random::<u64>()).collect())
@@ -305,7 +314,7 @@ mod tests {
         ];
 
         let without_mm = cfg
-            .compute_seq_hashes_for_tracking(&tokens, 2, None, BlockHashOptions::default())
+            .compute_seq_hashes_for_tracking(&tokens, 2, None, BlockHashOptions::default(), None)
             .unwrap();
         let with_mm = cfg
             .compute_seq_hashes_for_tracking(
@@ -316,9 +325,27 @@ mod tests {
                     block_mm_infos: Some(&mm_infos),
                     ..Default::default()
                 },
+                None,
             )
             .unwrap();
 
         assert_ne!(without_mm, with_mm);
     }
+
+    #[test]
+    fn compute_seq_hashes_for_tracking_uses_precomputed_block_hashes() {
+        let config = KvRouterConfig::default();
+        let tokens: Vec<u32> = (0..8).collect();
+        let precomputed = vec![LocalBlockHash(11), LocalBlockHash(29)];
+
+        let seq_hashes = config.compute_seq_hashes_for_tracking(
+            &tokens,
+            4,
+            None,
+            BlockHashOptions::default(),
+            Some(&precomputed),
+        );
+
+        assert_eq!(seq_hashes, Some(compute_seq_hash_for_block(&precomputed)));
+    }
 }

lib/kv-router/src/scheduling/selector.rs

@@ -22,36 +22,52 @@ pub trait WorkerSelector<C: WorkerConfigLike> {
 }
 
 /// Helper function for softmax sampling.
-/// Returns a vec of workers: multiple if tied, single if sampled.
+/// Returns the selected worker and its logit.
 fn softmax_sample(
     logits: &HashMap<WorkerWithDpRank, f64>,
     temperature: f64,
-) -> Vec<WorkerWithDpRank> {
+) -> (WorkerWithDpRank, f64) {
+    let mut rng = rand::rng();
+    softmax_sample_with_sample(logits, temperature, rng.random())
+}
+
+fn softmax_sample_with_sample(
+    logits: &HashMap<WorkerWithDpRank, f64>,
+    temperature: f64,
+    sample: f64,
+) -> (WorkerWithDpRank, f64) {
     if logits.is_empty() {
         panic!("Empty logits for softmax sampling");
     }
 
-    // Guard: if temperature is 0, return all keys with the smallest logit value (ties)
+    // Guard: at zero temperature, return a minimum-logit worker directly.
     if temperature == 0.0 {
-        let min_logit = logits.values().fold(f64::INFINITY, |a, &b| a.min(b));
-
-        let min_keys: Vec<_> = logits
-            .iter()
-            .filter(|&(_, &v)| v == min_logit)
-            .map(|(k, _)| *k)
-            .collect();
+        let mut logit_iter = logits.iter();
+        let (first_key, first_logit) = logit_iter.next().unwrap();
+
+        let mut min_logit = first_logit;
+        let mut min_key = first_key;
+        for (key, logit) in logit_iter {
+            if logit < min_logit {
+                min_logit = logit;
+                min_key = key;
+            }
+        }
 
-        return min_keys;
+        return (*min_key, *min_logit);
     }
 
-    let keys: Vec<_> = logits.keys().copied().collect();
-    let values: Vec<_> = logits.values().copied().collect();
+    let entries: Vec<_> = logits
+        .iter()
+        .map(|(worker, logit)| (*worker, *logit))
+        .collect();
+    let values: Vec<_> = entries.iter().map(|(_, logit)| *logit).collect();
 
     let min_val = values.iter().fold(f64::INFINITY, |a, &b| a.min(b));
     let max_val = values.iter().fold(f64::NEG_INFINITY, |a, &b| a.max(b));
 
     let probabilities = if min_val == max_val {
-        vec![1.0 / keys.len() as f64; keys.len()]
+        vec![1.0 / entries.len() as f64; entries.len()]
     } else {
         // Fused normalize -> negate -> scale -> exp, then normalize probabilities
         let range = max_val - min_val;
@@ -63,19 +79,16 @@ fn softmax_sample(
         probs
     };
 
-    let mut rng = rand::rng();
-    let sample: f64 = rng.random();
-
     let mut cumsum = 0.0;
     for (i, &prob) in probabilities.iter().enumerate() {
         cumsum += prob;
         if sample <= cumsum {
-            return vec![keys[i]];
+            return entries[i];
         }
     }
 
     // Fallback to last key (shouldn't normally reach here)
-    vec![keys[keys.len() - 1]]
+    entries[entries.len() - 1]
 }
 
 /// Default implementation matching the Python _cost_function.
@@ -118,76 +131,92 @@ impl<C: WorkerConfigLike> WorkerSelector<C> for DefaultWorkerSelector {
         let decode_blocks = &request.decode_blocks;
         let prefill_tokens = &request.prefill_tokens;
 
-        let mut worker_logits = HashMap::new();
-
         let overlap_weight = request
             .router_config_override
             .as_ref()
             .and_then(|cfg| cfg.overlap_score_weight)
             .unwrap_or(self.kv_router_config.overlap_score_weight);
 
-        for (worker_id, config) in workers
-            .iter()
-            .filter(|(wid, _)| allowed_ids.is_none_or(|ids| ids.contains(wid)))
-        {
-            let data_parallel_size = config.data_parallel_size();
-            let data_parallel_start_rank = config.data_parallel_start_rank();
+        let temperature = request
+            .router_config_override
+            .as_ref()
+            .and_then(|cfg| cfg.router_temperature)
+            .unwrap_or(self.kv_router_config.router_temperature);
 
-            for dp_rank in data_parallel_start_rank..(data_parallel_start_rank + data_parallel_size)
-            {
-                let worker = WorkerWithDpRank::new(*worker_id, dp_rank);
+        let get_score = |worker: WorkerWithDpRank| -> f64 {
+            let overlap = *overlaps.get(&worker).unwrap_or(&0);
+            let prefill_token = *prefill_tokens.get(&worker).unwrap_or(&isl);
+            let potential_prefill_block = (prefill_token as f64) / (block_size as f64);
+            let decode_block = *decode_blocks
+                .get(&worker)
+                .unwrap_or(&(potential_prefill_block.floor() as usize))
+                as f64;
 
-                let overlap = *overlaps.get(&worker).unwrap_or(&0);
+            let logit = overlap_weight * potential_prefill_block + decode_block;
 
-                let prefill_token = *prefill_tokens.get(&worker).unwrap_or(&isl);
-                let potential_prefill_block = (prefill_token as f64) / (block_size as f64);
+            tracing::debug!(
+                "Formula for worker_id={} dp_rank={:?} with {overlap} cached blocks: {logit:.3} \
+                 = {overlap_weight:.1} * prefill_blocks + decode_blocks \
+                 = {overlap_weight:.1} * {potential_prefill_block:.3} + {decode_block:.3}",
+                worker.worker_id,
+                worker.dp_rank
+            );
 
-                let decode_block = *decode_blocks
-                    .get(&worker)
-                    .unwrap_or(&(potential_prefill_block.floor() as usize))
-                    as f64;
+            logit
+        };
 
-                let logit = overlap_weight * potential_prefill_block + decode_block;
+        let worker_iter = workers
+            .iter()
+            .filter(move |(wid, _)| allowed_ids.is_none_or(|ids| ids.contains(wid)))
+            .flat_map(|(worker_id, config)| {
+                let data_parallel_size = config.data_parallel_size();
+                let data_parallel_start_rank = config.data_parallel_start_rank();
+                (data_parallel_start_rank..(data_parallel_start_rank + data_parallel_size))
+                    .map(move |dp_rank| WorkerWithDpRank::new(*worker_id, dp_rank))
+            });
 
-                worker_logits.insert(worker, logit);
+        let (best_worker, best_logit) = if temperature == 0.0 {
+            let mut min_workers = Vec::new();
+            let mut min_score = f64::INFINITY;
+            for worker in worker_iter {
+                let score = get_score(worker);
+                if score < min_score {
+                    min_workers.clear();
+                    min_workers.push(worker);
+                    min_score = score;
+                } else if score == min_score {
+                    min_workers.push(worker);
+                }
+            }
 
+            if min_workers.len() > 1 {
                 tracing::debug!(
-                    "Formula for worker_id={} dp_rank={:?} with {overlap} cached blocks: {logit:.3} \
-                     = {overlap_weight:.1} * prefill_blocks + decode_blocks \
-                     = {overlap_weight:.1} * {potential_prefill_block:.3} + {decode_block:.3}",
-                    worker.worker_id,
-                    worker.dp_rank
+                    "Multiple workers tied with same logit, using tree size as tie-breaker"
                 );
-            }
-        }
-
-        let temperature = request
-            .router_config_override
-            .as_ref()
-            .and_then(|cfg| cfg.router_temperature)
-            .unwrap_or(self.kv_router_config.router_temperature);
-        let candidates = softmax_sample(&worker_logits, temperature);
-
-        let best_worker = if candidates.len() > 1 {
-            tracing::debug!(
-                "Multiple workers tied with same logit, using tree size as tie-breaker"
-            );
-            let tree_sizes: Vec<(usize, &WorkerWithDpRank)> = candidates
-                .iter()
-                .map(|w| (request.overlaps.tree_sizes.get(w).copied().unwrap_or(0), w))
-                .collect();
-
-            if tree_sizes.iter().all(|(s, _)| *s == tree_sizes[0].0) {
-                let idx = rand::rng().random_range(0..candidates.len());
-                candidates[idx]
+                let tree_sizes: Vec<(usize, &WorkerWithDpRank)> = min_workers
+                    .iter()
+                    .map(|w| (request.overlaps.tree_sizes.get(w).copied().unwrap_or(0), w))
+                    .collect();
+
+                if tree_sizes.iter().all(|(s, _)| *s == tree_sizes[0].0) {
+                    let idx = rand::rng().random_range(0..min_workers.len());
+                    (min_workers[idx], min_score)
+                } else {
+                    let (_, worker) = *tree_sizes.iter().min_by_key(|(s, _)| *s).unwrap();
+                    (*worker, min_score)
+                }
             } else {
-                *tree_sizes.iter().min_by_key(|(s, _)| *s).unwrap().1
+                (min_workers[0], min_score)
             }
         } else {
-            candidates[0]
-        };
+            let mut worker_logits = HashMap::new();
+            for worker in worker_iter {
+                let score = get_score(worker);
+                worker_logits.insert(worker, score);
+            }
 
-        let best_logit = worker_logits[&best_worker];
+            softmax_sample(&worker_logits, temperature)
+        };
 
         if self.worker_type == "decode" {
             tracing::info!(
@@ -246,31 +275,22 @@ mod tests {
     fn test_softmax_sample_single_key() {
         let mut logits = HashMap::new();
         let worker = WorkerWithDpRank::from_worker_id(42);
-        logits.insert(worker, 0.5);
-
-        for temperature in &[0.1, 1.0, 10.0] {
-            let result = softmax_sample(&logits, *temperature);
-            assert_eq!(result.len(), 1, "Should return exactly one worker");
-            assert_eq!(result[0], worker, "Should return the only available worker");
+        for (logit, temperature) in [
+            (0.5, 0.1),
+            (0.5, 1.0),
+            (0.5, 10.0),
+            (-100.0, 1.0),
+            (100.0, 1.0),
+            (0.0, 1.0),
+            (0.0, 0.0),
+        ] {
+            logits.clear();
+            logits.insert(worker, logit);
+
+            let result = softmax_sample(&logits, temperature);
+            assert_eq!(result.0, worker, "Should return the only available worker");
+            assert_eq!(result.1, logit, "Should return the selected worker's logit");
         }
-
-        logits.clear();
-        logits.insert(worker, -100.0);
-        let result = softmax_sample(&logits, 1.0);
-        assert_eq!(result.len(), 1);
-        assert_eq!(result[0], worker);
-
-        logits.clear();
-        logits.insert(worker, 100.0);
-        let result = softmax_sample(&logits, 1.0);
-        assert_eq!(result.len(), 1);
-        assert_eq!(result[0], worker);
-
-        logits.clear();
-        logits.insert(worker, 0.0);
-        let result = softmax_sample(&logits, 1.0);
-        assert_eq!(result.len(), 1);
-        assert_eq!(result[0], worker);
     }
 
     #[test]
@@ -287,13 +307,12 @@ mod tests {
 
         let result = softmax_sample(&logits, 0.0);
         assert_eq!(
-            result.len(),
-            1,
-            "Should return one worker when there's no tie"
+            result.0, worker2,
+            "Should return worker with smallest logit when temperature is 0"
         );
         assert_eq!(
-            result[0], worker2,
-            "Should return worker with smallest logit when temperature is 0"
+            result.1, 3.0,
+            "Should return the smallest logit when temperature is 0"
         );
 
         logits.clear();
@@ -305,15 +324,11 @@ mod tests {
         logits.insert(worker6, 7.0);
 
         let result = softmax_sample(&logits, 0.0);
-        assert_eq!(
-            result.len(),
-            2,
-            "Should return all workers with smallest logit when tied"
-        );
         assert!(
-            result.contains(&worker2) && result.contains(&worker5),
-            "Should contain both tied workers"
+            result.0 == worker2 || result.0 == worker5,
+            "Should return one of the workers tied for the smallest logit"
         );
+        assert_eq!(result.1, 3.0, "Should return the tied minimum logit");
 
         logits.clear();
         let worker10 = WorkerWithDpRank::from_worker_id(10);
@@ -324,10 +339,44 @@ mod tests {
         logits.insert(worker30, 0.0);
 
         let result = softmax_sample(&logits, 0.0);
-        assert_eq!(result.len(), 1);
         assert_eq!(
-            result[0], worker20,
+            result.0, worker20,
             "Should handle negative logits correctly"
         );
+        assert_eq!(result.1, -5.0, "Should return the minimum negative logit");
+    }
+
+    #[test]
+    fn test_softmax_sample_with_sample_returns_selected_logit() {
+        let worker1 = WorkerWithDpRank::from_worker_id(1);
+        let worker2 = WorkerWithDpRank::from_worker_id(2);
+        let worker3 = WorkerWithDpRank::from_worker_id(3);
+
+        let logits = HashMap::from([(worker1, 0.0), (worker2, 3.0), (worker3, 9.0)]);
+        let entries: Vec<_> = logits
+            .iter()
+            .map(|(worker, logit)| (*worker, *logit))
+            .collect();
+        let values: Vec<_> = entries.iter().map(|(_, logit)| *logit).collect();
+
+        let min_val = values.iter().fold(f64::INFINITY, |a, &b| a.min(b));
+        let max_val = values.iter().fold(f64::NEG_INFINITY, |a, &b| a.max(b));
+        let temperature = 1.0;
+        let range = max_val - min_val;
+        let scaled: Vec<f64> = values.iter().map(|&v| -(v / range) / temperature).collect();
+        let max_scaled = scaled.iter().fold(f64::NEG_INFINITY, |a, &b| a.max(b));
+        let mut probabilities: Vec<f64> = scaled.iter().map(|&v| (v - max_scaled).exp()).collect();
+        let sum: f64 = probabilities.iter().sum();
+        probabilities.iter_mut().for_each(|p| *p /= sum);
+
+        let target_idx = entries
+            .iter()
+            .position(|(_, logit)| *logit > min_val)
+            .expect("expected at least one non-minimum logit");
+        let cumsum_before: f64 = probabilities.iter().take(target_idx).sum();
+        let sample = cumsum_before + probabilities[target_idx] / 2.0;
+
+        let result = softmax_sample_with_sample(&logits, temperature, sample);
+        assert_eq!(result, entries[target_idx]);
     }
 }

lib/llm/src/kv_router.rs

@@ -429,19 +429,26 @@ where
         }
 
         let isl_tokens = tokens.len();
+        let hash_options = BlockHashOptions {
+            block_mm_infos,
+            lora_name: lora_name.as_deref(),
+            is_eagle: Some(self.is_eagle),
+        };
 
-        let block_hashes = tracing::info_span!("kv_router.compute_block_hashes").in_scope(|| {
-            compute_block_hash_for_seq(
+        let block_hashes = tracing::info_span!("kv_router.compute_block_hashes")
+            .in_scope(|| compute_block_hash_for_seq(tokens, self.block_size, hash_options));
+        let hash_elapsed = start.elapsed();
+        // Compute seq_hashes only if scheduler needs it for active blocks tracking
+        let maybe_seq_hashes = tracing::info_span!("kv_router.compute_seq_hashes").in_scope(|| {
+            self.kv_router_config.compute_seq_hashes_for_tracking(
                 tokens,
                 self.block_size,
-                BlockHashOptions {
-                    block_mm_infos,
-                    lora_name: lora_name.as_deref(),
-                    is_eagle: Some(self.is_eagle),
-                },
+                router_config_override,
+                hash_options,
+                Some(&block_hashes),
             )
         });
-        let hash_elapsed = start.elapsed();
+        let seq_hash_elapsed = start.elapsed();
 
         let overlap_scores = self
             .indexer
@@ -450,21 +457,6 @@ where
             .await?;
         let find_matches_elapsed = start.elapsed();
 
-        // Compute seq_hashes only if scheduler needs it for active blocks tracking
-        let maybe_seq_hashes = tracing::info_span!("kv_router.compute_seq_hashes").in_scope(|| {
-            self.kv_router_config.compute_seq_hashes_for_tracking(
-                tokens,
-                self.block_size,
-                router_config_override,
-                BlockHashOptions {
-                    block_mm_infos,
-                    lora_name: lora_name.as_deref(),
-                    is_eagle: Some(self.is_eagle),
-                },
-            )
-        });
-        let seq_hash_elapsed = start.elapsed();
-
         let response = self
             .scheduler
             .schedule(
@@ -486,8 +478,8 @@ where
         if let Some(m) = metrics::RoutingOverheadMetrics::get() {
             m.observe(
                 hash_elapsed,
-                find_matches_elapsed,
                 seq_hash_elapsed,
+                find_matches_elapsed,
                 total_elapsed,
             );
         }
@@ -496,9 +488,9 @@ where
         tracing::info!(
             isl_tokens,
             hash_us = hash_elapsed.as_micros() as u64,
-            find_matches_us = (find_matches_elapsed - hash_elapsed).as_micros() as u64,
-            seq_hash_us = (seq_hash_elapsed - find_matches_elapsed).as_micros() as u64,
-            schedule_us = (total_elapsed - seq_hash_elapsed).as_micros() as u64,
+            seq_hash_us = (seq_hash_elapsed - hash_elapsed).as_micros() as u64,
+            find_matches_us = (find_matches_elapsed - seq_hash_elapsed).as_micros() as u64,
+            schedule_us = (total_elapsed - find_matches_elapsed).as_micros() as u64,
             total_us = total_elapsed.as_micros() as u64,
             "find_best_match completed"
         );
@@ -524,16 +516,18 @@ where
         router_config_override: Option<&RouterConfigOverride>,
     ) {
         let isl_tokens = tokens.len();
+        let hash_options = BlockHashOptions {
+            block_mm_infos,
+            lora_name: lora_name.as_deref(),
+            is_eagle: Some(self.is_eagle),
+        };
 
         let maybe_seq_hashes = self.kv_router_config.compute_seq_hashes_for_tracking(
             tokens,
             self.block_size,
             router_config_override,
-            BlockHashOptions {
-                block_mm_infos,
-                lora_name: lora_name.as_deref(),
-                is_eagle: Some(self.is_eagle),
-            },
+            hash_options,
+            None,
         );
 
         if let Err(e) = self
@@ -615,28 +609,23 @@ where
         lora_name: Option<&str>,
     ) -> Result<Vec<PotentialLoad>> {
         let isl_tokens = tokens.len();
-        let block_hashes = compute_block_hash_for_seq(
-            tokens,
-            self.block_size,
-            BlockHashOptions {
-                block_mm_infos,
-                lora_name,
-                is_eagle: Some(self.is_eagle),
-            },
-        );
-        let overlap_scores = self.indexer.find_matches(block_hashes.clone()).await?;
+        let hash_options = BlockHashOptions {
+            block_mm_infos,
+            lora_name,
+            is_eagle: Some(self.is_eagle),
+        };
+        let block_hashes = compute_block_hash_for_seq(tokens, self.block_size, hash_options);
 
         let maybe_seq_hashes = self.kv_router_config.compute_seq_hashes_for_tracking(
             tokens,
             self.block_size,
             router_config_override,
-            BlockHashOptions {
-                block_mm_infos,
-                lora_name,
-                is_eagle: Some(self.is_eagle),
-            },
+            hash_options,
+            Some(&block_hashes),
         );
 
+        let overlap_scores = self.indexer.find_matches(block_hashes).await?;
+
         Ok(self
             .scheduler
             .get_potential_loads(maybe_seq_hashes, isl_tokens, overlap_scores))

lib/llm/src/kv_router/metrics.rs

@@ -263,26 +263,26 @@ impl RoutingOverheadMetrics {
     pub fn observe(
         &self,
         hash_elapsed: Duration,
-        find_matches_elapsed: Duration,
         seq_hash_elapsed: Duration,
+        find_matches_elapsed: Duration,
         total_elapsed: Duration,
     ) {
         self.block_hashing
             .observe(hash_elapsed.as_secs_f64() * 1000.0);
+        self.seq_hashing
+            .observe(seq_hash_elapsed.saturating_sub(hash_elapsed).as_secs_f64() * 1000.0);
         self.indexer_find_matches.observe(
             find_matches_elapsed
-                .saturating_sub(hash_elapsed)
+                .saturating_sub(seq_hash_elapsed)
                 .as_secs_f64()
                 * 1000.0,
         );
-        self.seq_hashing.observe(
-            seq_hash_elapsed
+        self.scheduling.observe(
+            total_elapsed
                 .saturating_sub(find_matches_elapsed)
                 .as_secs_f64()
                 * 1000.0,
         );
-        self.scheduling
-            .observe(total_elapsed.saturating_sub(seq_hash_elapsed).as_secs_f64() * 1000.0);
         self.total.observe(total_elapsed.as_secs_f64() * 1000.0);
     }
 }
@@ -557,7 +557,7 @@ dynamo_frontend_router_queue_pending_requests{worker_type=\"decode\"} 5
             total: make("test_total_ms"),
         };
 
-        // Out-of-order durations: each phase < previous (would panic without saturating_sub)
+        // Out-of-order cumulative durations: each phase < previous (would panic without saturating_sub)
         metrics.observe(
             Duration::from_millis(10),
             Duration::from_millis(5),

lib/llm/src/kv_router/worker_query.rs

@@ -518,8 +518,8 @@ impl WorkerQueryClient {
                     events.len(),
                     last_event_id
                 );
-                for event in &events {
-                    self.indexer.apply_event(event.clone()).await;
+                for event in events {
+                    self.indexer.apply_event(event).await;
                 }
                 new_cursor = new_cursor.advance_to(last_event_id);
                 successful_response = true;

lib/mocker/src/replay/router/offline.rs

@@ -348,6 +348,7 @@ impl OfflineReplayRouter {
                         self.block_size,
                         None,
                         BlockHashOptions::default(),
+                        None,
                     )
                 };
                 (overlaps, token_seq)
@@ -359,6 +360,7 @@ impl OfflineReplayRouter {
                     self.block_size,
                     None,
                     BlockHashOptions::default(),
+                    None,
                 );
                 (overlaps, token_seq)
             }

lib/mocker/src/replay/router/online.rs

@@ -188,6 +188,7 @@ impl KvReplayRouter {
             self.block_size,
             None,
             BlockHashOptions::default(),
+            None,
         );
         let response = self
             .scheduler