[misc] improve memory profiling (vllm-project#11809)

Signed-off-by: youkaichao <[email protected]>
Co-authored-by: Cyrus Leung <[email protected]>

tests/test_utils.py

@@ -5,6 +5,7 @@
 
 import pytest
 import torch
+from vllm_test_utils import monitor
 
 from vllm.utils import (FlexibleArgumentParser, StoreBoolean, deprecate_kwargs,
                         get_open_port, memory_profiling, merge_async_iterators,
@@ -289,16 +290,32 @@ def test_memory_profiling():
 
     weights_memory_in_bytes = 128 * 1024 * 1024 * 4 # 512 MiB
 
+    def measure_current_non_torch():
+        free, total = torch.cuda.mem_get_info()
+        current_used = total - free
+        current_torch = torch.cuda.memory_reserved()
+        current_non_torch = current_used - current_torch
+        return current_non_torch
+
     with memory_profiling(baseline_memory_in_bytes=baseline_memory_in_bytes,
-    weights_memory_in_bytes=weights_memory_in_bytes) as result:
+    weights_memory_in_bytes=weights_memory_in_bytes) as result, \
+        monitor(measure_current_non_torch) as monitored_values:
         # make a memory spike, 1 GiB
         spike = torch.randn(256, 1024, 1024, device='cuda', dtype=torch.float32)
         del spike
 
         # Add some extra non-torch memory 256 MiB (simulate NCCL)
         handle2 = lib.cudaMalloc(256 * 1024 * 1024)
 
+    # this is an analytic value, it is exact,
+    # we only have 256 MiB non-torch memory increase
+    measured_diff = monitored_values.values[-1] - monitored_values.values[0]
+    assert measured_diff == 256 * 1024 * 1024
+
     # Check that the memory usage is within 5% of the expected values
+    # 5% tolerance is caused by PyTorch caching allocator,
+    # we cannot control PyTorch's behavior of its internal buffers,
+    # which causes a small error (<10 MiB in practice)
     non_torch_ratio = result.non_torch_increase_in_bytes / (256 * 1024 * 1024) # noqa
     torch_peak_ratio = result.torch_peak_increase_in_bytes / (1024 * 1024 * 1024) # noqa
     assert abs(non_torch_ratio - 1) <= 0.05

tests/vllm_test_utils/vllm_test_utils/__init__.py

@@ -4,5 +4,6 @@
 """
 
 from .blame import BlameResult, blame
+from .monitor import MonitoredValues, monitor
 
-__all__ = ["blame", "BlameResult"]
+__all__ = ["blame", "BlameResult", "monitor", "MonitoredValues"]

tests/vllm_test_utils/vllm_test_utils/monitor.py

@@ -0,0 +1,68 @@
+import contextlib
+import dataclasses
+import sys
+import traceback
+from typing import Callable, Generator, Generic, TypeVar
+
+_T = TypeVar("_T")
+
+
+@dataclasses.dataclass
+class MonitoredValues(Generic[_T]):
+    values: list[_T] = dataclasses.field(default_factory=list)
+    trace_stacks: list[str] = dataclasses.field(default_factory=list)
+
+
+@contextlib.contextmanager
+def monitor(
+    measure_func: Callable[[],
+                           _T]) -> Generator[MonitoredValues[_T], None, None]:
+    """
+    Trace the function calls to continuously monitor the change of
+    a value.
+
+    Usage:
+
+    ```python
+
+    def measure_func():
+        ... # measure the current value
+        return current_value
+
+    with monitor(measure_func) as monitored_values:
+        # do something
+    
+        monitored_values.values # all changes of the values
+        monitored_values.trace_stacks # trace stacks of every change
+    ```
+    """
+    monitored_values = MonitoredValues[_T]()
+
+    def _trace_calls(frame, event, arg=None):
+        nonlocal monitored_values
+        if event in ['line']:
+            # triggered by every line of Python code.
+            # only Python functions will trigger it,
+            # c/cpp functions will not trigger it.
+            try:
+                # Temporarily disable the trace function
+                sys.settrace(None)
+                # do a measurement
+                current_value = measure_func()
+                if len(monitored_values.values
+                       ) == 0 or current_value != monitored_values.values[-1]:
+                    monitored_values.values.append(current_value)
+                    monitored_values.trace_stacks.append("".join(
+                        traceback.format_stack()))
+                # Re-enable the trace function
+                sys.settrace(_trace_calls)
+            except NameError:
+                # modules are deleted during shutdown
+                pass
+        return _trace_calls
+
+    try:
+        sys.settrace(_trace_calls)
+        yield monitored_values
+    finally:
+        sys.settrace(None)

vllm/utils.py

@@ -1742,10 +1742,10 @@ class MemorySnapshot:
     timestamp: float = 0.0
 
     def measure(self):
-        self.torch_peak_in_bytes = torch.cuda.memory_stats(
-        )["allocated_bytes.all.peak"]
-        self.torch_memory_in_bytes = torch.cuda.memory_stats(
-        )["allocated_bytes.all.current"]
+        self.torch_peak_in_bytes = torch.cuda.max_memory_reserved()
+        # torch.cuda.memory_reserved() is how many bytes
+        # PyTorch gets from cuda (by calling cudaMalloc, etc.)
+        self.torch_memory_in_bytes = torch.cuda.memory_reserved()
         self.timestamp = time.time()
 
     def __sub__(self, other: "MemorySnapshot") -> "MemorySnapshot":
@@ -1822,10 +1822,10 @@ def memory_profiling(
 
     The memory used for loading weights (a.) is directly given from the argument `weights_memory_in_bytes`.
 
-    The increase of ``torch.cuda.memory_stats()["allocated_bytes.all.peak"]` after profiling gives (b.).
+    The increase of `torch.cuda.memory_stats()["allocated_bytes.all.peak"]` after profiling gives (b.).
 
     (c.) is tricky. We measure the total memory used in this GPU (`torch.cuda.mem_get_info()[1] - torch.cuda.mem_get_info()[0]`),
-    subtract the baseline memory, the memory used by the model weights, and diff of `torch.cuda.memory_stats()["allocated_bytes.all.current"]`.
+    subtract the baseline memory, the memory used by the model weights, and diff of `torch.cuda.memory_reserved()`.
     """ # noqa
     torch.cuda.reset_peak_memory_stats()