[torch.compile] transparent compilation with more logging (#12246)

Signed-off-by: youkaichao <[email protected]>

vllm/compilation/backends.py

@@ -524,6 +524,7 @@ def configure_post_pass(self):
 
     def __call__(self, graph: fx.GraphModule, example_inputs) -> Callable:
 
+        vllm_config = self.vllm_config
         if not self.compilation_config.cache_dir:
             # no provided cache dir, generate one based on the known factors
             # that affects the compilation. if none of the factors change,
@@ -532,7 +533,6 @@ def __call__(self, graph: fx.GraphModule, example_inputs) -> Callable:
 
             # 1. factors come from the vllm_config (it mainly summarizes how the
             #    model is created)
-            vllm_config = self.vllm_config
             config_hash = vllm_config.compute_hash()
 
             # 2. factors come from the code files that are traced by Dynamo (
@@ -556,20 +556,26 @@ def __call__(self, graph: fx.GraphModule, example_inputs) -> Callable:
             hash_key = hashlib.md5(
                 f"{config_hash}_{code_hash}".encode()).hexdigest()[:10]
             cache_dir = os.path.join(
-                envs.VLLM_CACHE_ROOT, "torch_compile_cache", hash_key,
-                f"rank_{vllm_config.parallel_config.rank}")
-        else:
-            cache_dir = self.compilation_config.cache_dir
+                envs.VLLM_CACHE_ROOT,
+                "torch_compile_cache",
+                hash_key,
+            )
+            self.compilation_config.cache_dir = cache_dir
+
+        cache_dir = self.compilation_config.cache_dir
         os.makedirs(cache_dir, exist_ok=True)
+        local_cache_dir = os.path.join(
+            cache_dir, f"rank_{vllm_config.parallel_config.rank}")
+        self.compilation_config.local_cache_dir = local_cache_dir
 
         disabled = envs.VLLM_DISABLE_COMPILE_CACHE
         self.inductor_hash_cache: InductorHashCache = InductorHashCache(
-            cache_dir, disabled=disabled)
+            local_cache_dir, disabled=disabled)
         if disabled:
             logger.info("vLLM's torch.compile cache is disabled.")
         else:
             logger.info("Using cache directory: %s for vLLM's torch.compile",
-                        cache_dir)
+                        local_cache_dir)
 
         # when dynamo calls the backend, it means the bytecode
         # transform and analysis are done
@@ -609,6 +615,18 @@ def __call__(self, graph: fx.GraphModule, example_inputs) -> Callable:
                                     self.vllm_config, self.graph_pool,
                                     self).run(*example_inputs)
 
+        graph_path = os.path.join(local_cache_dir, "computation_graph.py")
+        if not os.path.exists(graph_path):
+            # code adapted from https://github.com/thuml/depyf/blob/dab831108a752d1facc00acdd6d4243891845c37/depyf/explain/patched_lazy_format_graph_code.py#L30 # noqa
+            # use `print_readable` because it can include submodules
+            src = "from __future__ import annotations\nimport torch\n" + \
+                self.split_gm.print_readable(print_output=False)
+            src = src.replace("<lambda>", "GraphModule")
+            with open(graph_path, "w") as f:
+                f.write(src)
+
+            logger.debug("Computation graph saved to %s", graph_path)
+
         self._called = True
 
         if not self.compilation_config.use_cudagraph or \

vllm/compilation/decorators.py

@@ -198,6 +198,8 @@ def __call__(self, *args, **kwargs):
                             f" {dims} for argument {k} with type {type(arg)}.")
             # here, it is the starting point of the `torch.compile` process
             start_monitoring_torch_compile(self.vllm_config)
+            logger.debug("Start compiling function %s",
+                         self.original_code_object)
 
         # if we don't use custom dispatcher, we can directly call the
         # compiled function and let torch.compile handle the dispatching,

vllm/compilation/wrapper.py

@@ -9,6 +9,9 @@
 
 import vllm.envs as envs
 from vllm.config import CompilationLevel, get_current_vllm_config
+from vllm.logger import init_logger
+
+logger = init_logger(__name__)
 
 
 class TorchCompileWrapperWithCustomDispatcher:
@@ -82,6 +85,25 @@ def bytecode_hook(self, old_code: CodeType, new_code: CodeType):
             return
 
         self.compiled_codes.append(new_code)
+        local_cache_dir = self.vllm_config.compilation_config.local_cache_dir
+        if isinstance(local_cache_dir, str):
+            decompiled_file = os.path.join(local_cache_dir,
+                                           "transformed_code.py")
+            if not os.path.exists(decompiled_file):
+                try:
+                    # usually the decompilation will succeed for most models,
+                    # as we guarantee a full-graph compilation in Dynamo.
+                    # but there's no 100% guarantee, since decompliation is
+                    # not a reversible process.
+                    import depyf
+                    src = depyf.decompile(new_code)
+                    with open(decompiled_file, "w") as f:
+                        f.write(src)
+
+                    logger.debug("Dynamo transformed code saved to %s",
+                                 decompiled_file)
+                except Exception:
+                    pass
 
         if self.vllm_config.compilation_config.use_cudagraph and \
             "update" in new_code.co_names:

vllm/config.py

@@ -2785,6 +2785,7 @@ def model_post_init(self, __context: Any) -> None:
     compile_sizes: List[int] = PrivateAttr
     capture_sizes: List[int] = PrivateAttr
     max_capture_size: int = PrivateAttr
+    local_cache_dir: str = PrivateAttr  # local cache dir for each rank
     # optimization:
     # Intuitively, bs_to_padded_graph_size should be Dict[int, int].
     # since we know all keys are in a range [0, max_capture_size],