From a437373ff7f6a6f2070d7703738aad6634dd5338 Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Mon, 4 May 2026 16:32:14 +0000
Subject: [PATCH] =?UTF-8?q?=E2=9A=A1=20Bolt:=20Replace=20O(N^2)=20list=20p?=
 =?UTF-8?q?opping=20with=20O(N)=20index=20tracking=20in=20entropy=5Futils.?=
 =?UTF-8?q?py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Replaces `list.pop(0)` within nested loops with an O(1) index increment
variable `entropy_idx` to iterate over elements sequentially. This
improves execution time inside `calculate_logits_entropy` and
`speculate_calculate_logits_entropy` from O(N^2) to O(N).
---
 .jules/bolt.md                             | 3 +++
 fastdeploy/model_executor/entropy_utils.py | 8 ++++++--
 2 files changed, 9 insertions(+), 2 deletions(-)
 create mode 100644 .jules/bolt.md

diff --git a/.jules/bolt.md b/.jules/bolt.md
new file mode 100644
index 00000000000..d2508be30ba
--- /dev/null
+++ b/.jules/bolt.md
@@ -0,0 +1,3 @@
+## 2024-05-04 - Inefficient list.pop(0) operations
+**Learning:** Found multiple usages of `list.pop(0)` in the codebase, particularly in `fastdeploy/model_executor/entropy_utils.py` . `pop(0)` on a list is an O(N) operation because it requires shifting all subsequent elements. For queues or iterating over elements sequentially, using an index or `collections.deque` is much faster.
+**Action:** Replace O(N) `pop(0)` operations with O(1) index tracking or `collections.deque` where appropriate to optimize execution performance.
diff --git a/fastdeploy/model_executor/entropy_utils.py b/fastdeploy/model_executor/entropy_utils.py
index 21d1b3421e9..e97a38d39a9 100644
--- a/fastdeploy/model_executor/entropy_utils.py
+++ b/fastdeploy/model_executor/entropy_utils.py
@@ -49,9 +49,11 @@ def calculate_logits_entropy(logits, share_inputs, temperature):
     entropy_tensor = get_entropy(logits)
     entropy = entropy_tensor.tolist()
 
+    entropy_idx = 0
     for i in range(real_bsz):
         for _ in range(real_seq_lens[i]):
-            share_inputs["entropy_list"][i].append(entropy.pop(0))
+            share_inputs["entropy_list"][i].append(entropy[entropy_idx])
+            entropy_idx += 1
         if (
             share_inputs["stop_flags"][i]
             and share_inputs["seq_lens_decoder"][i] != 0
@@ -92,9 +94,11 @@ def speculate_calculate_logits_entropy(logits, share_inputs, temperature):
     entropy_tensor = get_entropy(accepted_logits)
     entropy = entropy_tensor.tolist()
 
+    entropy_idx = 0
     for i in range(real_bsz):
         for _ in range(share_inputs["accept_num"][i]):
-            share_inputs["entropy_list"][i].append(entropy.pop(0))
+            share_inputs["entropy_list"][i].append(entropy[entropy_idx])
+            entropy_idx += 1
         if (
             share_inputs["stop_flags"][i]
             and share_inputs["seq_lens_decoder"][i] != 0