perf(engine): reduce proof worker count for small blocks (#22074)

Co-authored-by: Amp <amp@ampcode.com>
Co-authored-by: Ubuntu <ubuntu@dev-yk.tail388b2e.ts.net>
Co-authored-by: YK <chiayongkang@hotmail.com>

crates/engine/tree/src/tree/payload_processor/mod.rs

@@ -97,6 +97,7 @@ pub const SPARSE_TRIE_MAX_VALUES_SHRINK_CAPACITY: usize = 1_000_000;
 /// Blocks with fewer transactions than this skip prewarming, since the fixed overhead of spawning
 /// prewarm workers exceeds the execution time saved.
 pub const SMALL_BLOCK_TX_THRESHOLD: usize = 5;
+
 /// Type alias for [`PayloadHandle`] returned by payload processor spawn methods.
 type IteratorPayloadHandle<Evm, I, N> = PayloadHandle<
     WithTxEnv<TxEnvFor<Evm>, <I as ExecutableTxIterator<Evm>>::Recovered>,
@@ -251,8 +252,9 @@ where
         // Extract V2 proofs flag early so we can pass it to prewarm
         let v2_proofs_enabled = !config.disable_proof_v2();
 
-        // Capture parent_state_root before env is moved into spawn_caching_with
+        // Capture fields before env is moved into spawn_caching_with
         let parent_state_root = env.parent_state_root;
+        let transaction_count = env.transaction_count;
 
         // Handle BAL-based optimization if available
         let prewarm_handle = if let Some(bal) = bal {
@@ -281,9 +283,11 @@ where
             )
         };
 
-        // Create and spawn the storage proof task
+        // Create and spawn the storage proof task.
         let task_ctx = ProofTaskCtx::new(multiproof_provider_factory);
-        let proof_handle = ProofWorkerHandle::new(&self.executor, task_ctx, v2_proofs_enabled);
+        let halve_workers = transaction_count <= Self::SMALL_BLOCK_PROOF_WORKER_TX_THRESHOLD;
+        let proof_handle =
+            ProofWorkerHandle::new(&self.executor, task_ctx, halve_workers, v2_proofs_enabled);
 
         if config.disable_trie_cache() {
             let multi_proof_task = MultiProofTask::new(
@@ -363,6 +367,10 @@ where
         }
     }
 
+    /// Transaction count threshold below which proof workers are halved, since fewer transactions
+    /// produce fewer state changes and most workers would be idle overhead.
+    const SMALL_BLOCK_PROOF_WORKER_TX_THRESHOLD: usize = 30;
+
     /// Transaction count threshold below which sequential signature recovery is used.
     ///
     /// For blocks with fewer than this many transactions, the rayon parallel iterator overhead

crates/engine/tree/src/tree/payload_processor/multiproof.rs

@@ -1582,7 +1582,7 @@ mod tests {
         let changeset_cache = ChangesetCache::new();
         let overlay_factory = OverlayStateProviderFactory::new(factory, changeset_cache);
         let task_ctx = ProofTaskCtx::new(overlay_factory);
-        let proof_handle = ProofWorkerHandle::new(runtime, task_ctx, false);
+        let proof_handle = ProofWorkerHandle::new(runtime, task_ctx, false, false);
         let (to_sparse_trie, _receiver) = std::sync::mpsc::channel();
         let (tx, rx) = crossbeam_channel::unbounded();