From 7906d1d29863bc3b33c4bcfb942a5d61f9867127 Mon Sep 17 00:00:00 2001
From: Lianmin Zheng <lianminzheng@gmail.com>
Date: Sat, 18 Jan 2025 20:20:23 -0800
Subject: [PATCH] Remove the unused write_with_records (#2972)

---
 python/sglang/srt/managers/schedule_batch.py  |  1 -
 python/sglang/srt/mem_cache/memory_pool.py    | 28 +------------------
 .../sglang/srt/model_executor/model_runner.py |  1 -
 3 files changed, 1 insertion(+), 29 deletions(-)

diff --git a/python/sglang/srt/managers/schedule_batch.py b/python/sglang/srt/managers/schedule_batch.py
index cec2262c4..afbc98b7c 100644
--- a/python/sglang/srt/managers/schedule_batch.py
+++ b/python/sglang/srt/managers/schedule_batch.py
@@ -158,7 +158,6 @@ class ImageInputs:
     im_end_id: Optional[torch.Tensor] = None
     slice_start_id: Optional[torch.Tensor] = None
     slice_end_id: Optional[torch.Tensor] = None
-
     tgt_sizes: Optional[list] = None
 
     @staticmethod
diff --git a/python/sglang/srt/mem_cache/memory_pool.py b/python/sglang/srt/mem_cache/memory_pool.py
index ab27e81b7..e30736722 100644
--- a/python/sglang/srt/mem_cache/memory_pool.py
+++ b/python/sglang/srt/mem_cache/memory_pool.py
@@ -49,7 +49,6 @@ class ReqToTokenPool:
         size: int,
         max_context_len: int,
         device: str,
-        use_records: bool,
         enable_memory_saver: bool,
     ):
         memory_saver_adapter = TorchMemorySaverAdapter.create(
@@ -64,17 +63,9 @@ class ReqToTokenPool:
                 (size, max_context_len), dtype=torch.int32, device=device
             )
         self.free_slots = list(range(size))
-        self.write_records = []
-        self.use_records = use_records
-
-        if self.use_records:
-            self.write = self.write_with_records
-        else:
-            self.write = self.write_without_records
 
     def write(self, indices, values):
-        # Keep the signature for type checking. It will be assigned during runtime.
-        raise NotImplementedError()
+        self.req_to_token[indices] = values
 
     def available_size(self):
         return len(self.free_slots)
@@ -96,23 +87,6 @@ class ReqToTokenPool:
 
     def clear(self):
         self.free_slots = list(range(self.size))
-        self.write_records = []
-
-    def write_without_records(self, indices, values):
-        self.req_to_token[indices] = values
-
-    def write_with_records(self, indices, values):
-        self.req_to_token[indices] = values
-        self.write_records.append((indices, values))
-
-    def get_write_records(self):
-        ret = self.write_records
-        self.write_records = []
-        return ret
-
-    def apply_write_records(self, write_records: List[Tuple]):
-        for indices, values in write_records:
-            self.req_to_token[indices] = values
 
 
 class BaseTokenToKVPool:
diff --git a/python/sglang/srt/model_executor/model_runner.py b/python/sglang/srt/model_executor/model_runner.py
index bca4711eb..46920d922 100644
--- a/python/sglang/srt/model_executor/model_runner.py
+++ b/python/sglang/srt/model_executor/model_runner.py
@@ -617,7 +617,6 @@ class ModelRunner:
             size=max_num_reqs + 1,
             max_context_len=self.model_config.context_len + 4,
             device=self.device,
-            use_records=False,
             enable_memory_saver=self.server_args.enable_memory_saver,
         )
         if (