Merge leftover

Author: gshtras

benchmarks/kernels/benchmark_paged_attention.py

@@ -169,7 +169,6 @@ def run_cuda_benchmark(num_iters: int, profile: bool = False) -> float:
                         kv_cache_dtype,
                         k_scale,
                         v_scale,
-                        None,
                     )
             else:
                 raise ValueError(f"Invalid version: {version}")
@@ -183,13 +182,13 @@ def run_cuda_benchmark(num_iters: int, profile: bool = False) -> float:
     # Warmup.
     print("Warming up...")
     run_benchmark = run_cuda_benchmark
-    run_benchmark(num_iters=500, profile=False)
+    run_benchmark(num_iters=3, profile=False)
 
     # Benchmark.
     if do_profile:
         latency = run_benchmark(num_iters=1, profile=True)
     else:
-        latency = run_benchmark(num_iters=10000, profile=False)
+        latency = run_benchmark(num_iters=100, profile=False)
     print(f"Kernel running time: {latency * 1000000:.3f} us")
 
 

vllm/_custom_ops.py

@@ -11,7 +11,6 @@
 from vllm.logger import init_logger
 from vllm.platforms import current_platform
 from vllm.scalar_type import ScalarType
-from vllm.utils import is_navi
 
 logger = init_logger(__name__)