diff --git a/bench_offload.py b/bench_offload.py
index 97224aa..33d46cf 100644
--- a/bench_offload.py
+++ b/bench_offload.py
@@ -5,7 +5,7 @@ from nanovllm import LLM, SamplingParams
 
 
 def bench_decode(llm, num_seqs, max_input_len, max_output_len):
-    """Benchmark decode performance"""
+    """Benchmark decode performance (original test)"""
     seed(0)
     prompt_token_ids = [[randint(0, 10000) for _ in range(randint(100, max_input_len))] for _ in range(num_seqs)]
     sampling_params = [SamplingParams(temperature=0.6, ignore_eos=True, max_tokens=randint(100, max_output_len)) for _ in range(num_seqs)]
@@ -21,6 +21,7 @@ def bench_decode(llm, num_seqs, max_input_len, max_output_len):
 def bench_prefill(llm, num_seqs, input_len):
     """Benchmark prefill performance"""
     seed(0)
+    # Fixed length input, minimal output to focus on prefill
     prompt_token_ids = [[randint(0, 10000) for _ in range(input_len)] for _ in range(num_seqs)]
     sampling_params = SamplingParams(temperature=0.6, ignore_eos=True, max_tokens=1)
 
@@ -40,6 +41,8 @@ def main():
         max_model_len=128 * 1024,
         max_num_batched_tokens=128 * 1024,
         enable_cpu_offload=True,
+        num_gpu_blocks=6,
+        num_prefetch_blocks=2,
     )
 
     # Warmup
@@ -48,15 +51,16 @@ def main():
     print("=" * 60)
     print("Prefill Benchmark (CPU Offload)")
     print("=" * 60)
-    bench_prefill(llm, num_seqs=1, input_len=64*1024)
-    # bench_prefill(llm, num_seqs=1, input_len=16384)
-    # bench_prefill(llm, num_seqs=1, input_len=32000)
+    # bench_prefill(llm, num_seqs=1, input_len=1024)
+    # bench_prefill(llm, num_seqs=1, input_len=2048)
+    # bench_prefill(llm, num_seqs=1, input_len=4096)
+    bench_prefill(llm, num_seqs=1, input_len=8192)
 
     print("=" * 60)
     print("Decode Benchmark (CPU Offload)")
     print("=" * 60)
-    bench_decode(llm, num_seqs=1, max_input_len=64*1024, max_output_len=256)
-    # bench_decode(llm, num_seqs=1, max_input_len=16384, max_output_len=256)
+    bench_decode(llm, num_seqs=1, max_input_len=1024, max_output_len=128)
+    # bench_decode(llm, num_seqs=1, max_input_len=2048, max_output_len=128)
 
 
 if __name__ == "__main__":
diff --git a/nanovllm/engine/model_runner.py b/nanovllm/engine/model_runner.py
index e7f3c8a..caab2ed 100644
--- a/nanovllm/engine/model_runner.py
+++ b/nanovllm/engine/model_runner.py
@@ -7,7 +7,7 @@ from multiprocessing.shared_memory import SharedMemory
 from nanovllm.config import Config
 from nanovllm.engine.sequence import Sequence
 from nanovllm.models.qwen3 import Qwen3ForCausalLM
-from nanovllm.layers.sampler import Sampler
+from nanovllm.layers.sampler import GreedySampler
 from nanovllm.utils.context import set_context, get_context, reset_context
 from nanovllm.utils.loader import load_model
 from nanovllm.utils.logger import get_logger
@@ -34,7 +34,7 @@ class ModelRunner:
         torch.set_default_device("cuda")
         self.model = Qwen3ForCausalLM(hf_config)
         load_model(self.model, config.model)
-        self.sampler = Sampler()
+        self.sampler = GreedySampler()
         self.warmup_model()
         self.allocate_kv_cache()
         if not self.enforce_eager:
diff --git a/nanovllm/kvcache/hybrid_manager.py b/nanovllm/kvcache/hybrid_manager.py
index 019c27b..cc02f8b 100644
--- a/nanovllm/kvcache/hybrid_manager.py
+++ b/nanovllm/kvcache/hybrid_manager.py
@@ -1039,6 +1039,8 @@ class HybridKVCacheManager(KVCacheManager):
         """
         assert not seq.block_table, "Sequence already has blocks"
 
+        h = -1  # Running hash for prefix cache
+
         for i in range(seq.num_blocks):
             # Allocate CPU block
             if not self.free_cpu_blocks:
@@ -1049,10 +1051,19 @@ class HybridKVCacheManager(KVCacheManager):
 
             cpu_block_id = self.free_cpu_blocks.popleft()
 
+            # Get token IDs for this block and compute hash
+            token_ids = seq.block(i)
+            if len(token_ids) == self._block_size:
+                h = self.compute_hash(token_ids, h)
+            else:
+                h = -1  # Incomplete block
+
             # Allocate logical block
             logical_id = self.free_logical_ids.popleft()
             block = self.logical_blocks[logical_id]
             block.ref_count = 1
+            block.hash = h
+            block.token_ids = token_ids.copy() if len(token_ids) == self._block_size else []
             block.location = BlockLocation.CPU
             block.cpu_block_id = cpu_block_id
             block.gpu_slot = -1
@@ -1060,6 +1071,10 @@ class HybridKVCacheManager(KVCacheManager):
             self.cpu_block_to_logical[cpu_block_id] = logical_id
             seq.block_table.append(logical_id)
 
+            # Update prefix cache
+            if h != -1:
+                self.hash_to_logical_id[h] = logical_id
+
     def get_cpu_block_table(self, seq: Sequence) -> List[int]:
         """
         Get CPU block ID list for sequence.
diff --git a/nanovllm/layers/sampler.py b/nanovllm/layers/sampler.py
index b101018..dacb978 100644
--- a/nanovllm/layers/sampler.py
+++ b/nanovllm/layers/sampler.py
@@ -13,3 +13,13 @@ class Sampler(nn.Module):
         probs = torch.softmax(logits, dim=-1)
         sample_tokens = probs.div_(torch.empty_like(probs).exponential_(1).clamp_min_(1e-10)).argmax(dim=-1)
         return sample_tokens
+
+
+class GreedySampler(nn.Module):
+
+    def __init__(self):
+        super().__init__()
+
+    @torch.compile
+    def forward(self, logits: torch.Tensor, temperatures: torch.Tensor = None):
+        return logits.argmax(dim=-1)