Add VLLM_USE_PA_PRINT_PARAM flag to print pa size

CMakeLists.txt

@@ -160,7 +160,7 @@ set(VLLM_EXT_SRC
   "csrc/attention/attention_kernels_opt.cu"
   "csrc/opt/layernorm_kernels_opt.cu"
   "csrc/quantization/squeezellm/quant_cuda_kernel.cu"
-  #"csrc/quantization/gptq/q_gemm.cu"
+  "csrc/quantization/gptq/q_gemm.cu"
   "csrc/quantization/compressed_tensors/int8_quant_kernels.cu"
   # "csrc/quantization/fp8/common.cu"
   "csrc/cuda_utils_kernels.cu"

vllm/attention/ops/paged_attn.py

@@ -123,6 +123,11 @@ class PagedAttention:
 
         if use_v1:
             # Run PagedAttention V1.
+            if envs.VLLM_USE_PA_PRINT_PARAM:
+                print("PA V1 SIZE:")
+                print(f"query.shape = {query.shape}, key_cache.shape = {key_cache.shape}, value_cache.shape = {value_cache.shape}")
+                print(f"num_kv_heads = {num_kv_heads}, scale = {scale:.3f}, block_tables.shape = {block_tables.shape}, seq_lens.shape = {seq_lens.shape}, block_size = {block_size}, max_seq_len = {max_seq_len}")
+                
             if envs.USE_VLLM_OPT_OP:
                 ops.paged_attention_v1_opt(
                     output,
@@ -179,6 +184,13 @@ class PagedAttention:
                 device=output.device,
             )
             max_logits = torch.empty_like(exp_sums)
+            
+            if envs.VLLM_USE_PA_PRINT_PARAM:
+                print("PA V2 SIZE:")
+                print(f"exp_sums.shape = {exp_sums.shape}, max_logits.shape = {max_logits.shape}, tmp_output.shape = {tmp_output.shape}")
+                print(f"query.shape = {query.shape}, key_cache.shape = {key_cache.shape}, value_cache.shape = {value_cache.shape}")
+                print(f"num_kv_heads = {num_kv_heads}, scale = {scale:.3f}, block_tables.shape = {block_tables.shape}, seq_lens.shape = {seq_lens.shape}, block_size = {block_size}, max_seq_len = {max_seq_len}")
+                
             if envs.USE_VLLM_OPT_OP:
                 ops.paged_attention_v2_opt(
                 output,

vllm/envs.py

@@ -11,6 +11,7 @@ if TYPE_CHECKING:
     VLLM_USE_TRITON_FLASH_ATTN: bool = False
     VLLM_USE_FLASH_ATTN_AUTO: bool = False
     USE_VLLM_OPT_OP: bool = False
+    VLLM_USE_PA_PRINT_PARAM: bool = False 
     LOCAL_RANK: int = 0
     CUDA_VISIBLE_DEVICES: Optional[str] = None
     VLLM_ENGINE_ITERATION_TIMEOUT_S: int = 60
@@ -144,6 +145,11 @@ environment_variables: Dict[str, Callable[[], Any]] = {
     "USE_VLLM_OPT_OP":
     lambda: (os.environ.get("USE_VLLM_OPT_OP", "True").lower() in
              ("true", "1")),
+    
+    # flag to control if vllm print pa parameters
+    "VLLM_USE_PA_PRINT_PARAM":
+    lambda: (os.environ.get("VLLM_USE_PA_PRINT_PARAM", "False").lower() in
+             ("true", "1")),
 
     # local rank of the process in the distributed setting, used to determine
     # the GPU device id