From 5323969fcfd5c88817e67cb8136af10ed5c70f6a Mon Sep 17 00:00:00 2001
From: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Date: Wed, 24 Apr 2024 08:56:58 +0000
Subject: [PATCH] Increase #blocks

---
 vllm/worker/tpu_worker.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/vllm/worker/tpu_worker.py b/vllm/worker/tpu_worker.py
index 0bd1b8437da24..d69acb4ac16bf 100644
--- a/vllm/worker/tpu_worker.py
+++ b/vllm/worker/tpu_worker.py
@@ -64,7 +64,7 @@ class TPUWorker(LoraNotSupportedWorkerBase):
         self.model_runner.load_model()
 
     def determine_num_available_blocks(self) -> Tuple[int, int]:
-        num_tpu_blocks = 100
+        num_tpu_blocks = 1000
         return num_tpu_blocks, 0
 
     def initialize_cache(