Fix hook count performance regression from v0.18.5 (#7886)

Fixes performance regressions reported in #7882 and #7885.

PR #7780 added dynamic hook count computation for reentrant
checkpointing correctness, but placed the call inside every gradient
hook closure. For a model with n parameter tensors, this creates
significant overhead per backward pass.

Summary:

1. Added `should_refresh_expected_hook_count()` predicate that returns
true only at backward phase boundaries (first hook, or new reentrant
phase), so `count_used_parameters_in_backward()` is called once per
phase instead of once per hook.
2. Applied this predicate in ZeRO-1/2 (stage_1_and_2.py) and both ZeRO-3
hook sites (stage3.py), reusing the `cached_max_expected_hooks_seen`
value when refresh isn't needed.
3. Changed enter_backward() to reset hook counters on first real
backward entry, preventing pollution from pre-user-backward autograd
calls (e.g., TiledFusedLogitsLoss).

With 24-layer transformer, ~267M params (147 parameter tensors), ZeRO-2,
8×H100 80GB, bf16, batch size 8, 20 warmup + 20 measured iterations:
  - Before fix: 0.1265s/iter
  - After fix: 0.0505s/iter

---------

Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com>
Co-authored-by: Ramya Ramineni <rraminen@users.noreply.github.com>
Signed-off-by: nathon-lee <leejianwoo@gmail.com>

Author: 2 people

deepspeed/runtime/base_optimizer.py

@@ -109,6 +109,17 @@ def run_grad_acc_post_hooks(self):
 
     def enter_backward(self):
         """Enter backward context. Call at the start of backward pass."""
+        # On first real backward entry of a step, reset counters that may have been
+        # polluted by pre-user-backward hooks (e.g. TiledFusedLogitsLoss calling
+        # torch.autograd.backward() from forward). Do NOT reset on reentrant
+        # phase re-entry (backward_seen_this_step == True) so phase-to-phase
+        # state remains intact.
+        if self.backward_active_depth == 0 and not self.backward_seen_this_step:
+            self.hooks_fired_this_backward = 0
+            self.max_expected_hooks_seen = 0
+            self.remaining_grad_acc_hooks = 0
+            self.post_backward_callback_queued = False
+            self.post_backward_callback_graph_task_id = None
         self.backward_active_depth += 1
         # Track that backward has been active at some point in this step.
         # This is used to detect subsequent gradient hook phases with reentrant checkpointing.
@@ -128,6 +139,22 @@ def reset_for_new_step(self):
         self.post_backward_callback_queued = False
         self.post_backward_callback_graph_task_id = None
 
+    def should_refresh_expected_hook_count(self):
+        """Return True when count_used_parameters_in_backward() should be re-evaluated.
+
+        Refresh is needed in two cases:
+        1. First hook of a backward (or backward phase): hooks_fired == 0.
+        2. A new reentrant phase started: remaining hooks exhausted, we exited
+           backward, but backward was active earlier this step.
+
+        The predicate must be evaluated BEFORE reenter_backward_if_needed()
+        because re-entering changes backward_active_depth and hides the
+        phase-boundary signal.
+        """
+        return (self.hooks_fired_this_backward == 0
+                or (self.remaining_grad_acc_hooks == 0 and self.backward_active_depth == 0
+                    and self.backward_seen_this_step))
+
     def reenter_backward_if_needed(self):
         """Re-enter backward context for subsequent phases in reentrant checkpointing.
 
@@ -401,6 +428,10 @@ def clear_backward_seen_flag(self):
         """Clear the backward seen flag and reset hook counters at the start of each step."""
         self._backward_hook_state.reset_for_new_step()
 
+    def should_refresh_expected_hook_count(self):
+        """Return True when count_used_parameters_in_backward() should be re-evaluated."""
+        return self._backward_hook_state.should_refresh_expected_hook_count()
+
     def reenter_backward_if_needed(self):
         """Re-enter backward context for subsequent phases in reentrant checkpointing."""
         self._backward_hook_state.reenter_backward_if_needed()

deepspeed/runtime/zero/stage3.py

@@ -1279,14 +1279,19 @@ def wrapper(param):
 
                         @instrument_w_nvtx
                         def reduce_partition_and_remove_grads(*notneeded):
+                            # Evaluate refresh condition before reenter_backward_if_needed()
+                            refresh_expected = self.should_refresh_expected_hook_count()
                             # Re-enter backward for subsequent phases in reentrant checkpointing
                             self.reenter_backward_if_needed()
 
                             self.reduce_ready_partitions_and_remove_grads(param)
 
                             # Update hook state and run epilogue if all expected hooks have fired
-                            current_expected = count_used_parameters_in_backward(
-                                non_leaf_params_requiring_grad) + leaf_module_count
+                            if refresh_expected:
+                                current_expected = count_used_parameters_in_backward(
+                                    non_leaf_params_requiring_grad) + leaf_module_count
+                            else:
+                                current_expected = self._max_expected_hooks_seen
                             self.update_hook_state_and_maybe_run_epilogue(current_expected)
 
                         self._grad_acc_hooks.append(register_grad_hook(param, reduce_partition_and_remove_grads))
@@ -1303,6 +1308,8 @@ def reduce_partition_and_remove_grads(*notneeded):
             def make_hook(params):
 
                 def reduce_leaf_module_grads(module, grad_input, grad_output):
+                    # Evaluate refresh condition before reenter_backward_if_needed()
+                    refresh_expected = self.should_refresh_expected_hook_count()
                     self.reenter_backward_if_needed()
 
                     for param in params:
@@ -1311,8 +1318,11 @@ def reduce_leaf_module_grads(module, grad_input, grad_output):
                             param.grad = torch.zeros_like(param)
                         self.reduce_ready_partitions_and_remove_grads(param)
 
-                    current_expected = count_used_parameters_in_backward(
-                        non_leaf_params_requiring_grad) + leaf_module_count
+                    if refresh_expected:
+                        current_expected = count_used_parameters_in_backward(
+                            non_leaf_params_requiring_grad) + leaf_module_count
+                    else:
+                        current_expected = self._max_expected_hooks_seen
                     self.update_hook_state_and_maybe_run_epilogue(current_expected)
 
                 return reduce_leaf_module_grads

deepspeed/runtime/zero/stage_1_and_2.py

@@ -1053,9 +1053,14 @@ def create_gradient_handling_hooks(self):
                     def wrapper(param, i):
 
                         def grad_handling_hook(*notneeded):
+                            # Evaluate refresh condition before reenter_backward_if_needed()
+                            refresh_expected = self.should_refresh_expected_hook_count()
                             self.reenter_backward_if_needed()
                             self.process_gradients(param, i)
-                            current_expected = count_used_parameters_in_backward(all_params_requiring_grad)
+                            if refresh_expected:
+                                current_expected = count_used_parameters_in_backward(all_params_requiring_grad)
+                            else:
+                                current_expected = self._max_expected_hooks_seen
                             self.update_hook_state_and_maybe_run_epilogue(current_expected)
 
                         self._grad_acc_hooks.append(register_grad_hook(param, grad_handling_hook))

tests/unit/v1/zero/test_zero_hook_count_regression.py

@@ -0,0 +1,104 @@
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+"""Regression tests for count_used_parameters_in_backward() call count.
+
+Verifies fix for https://github.com/deepspeedai/DeepSpeed/issues/7885:
+count_used_parameters_in_backward() was called once per gradient hook
+(O(n) calls per backward) instead of once per backward phase (O(1)
+for non-reentrant, O(p) for reentrant with p phases).
+"""
+
+import pytest
+import torch
+from unittest.mock import patch
+
+import deepspeed
+from deepspeed.accelerator import get_accelerator
+from unit.common import DistributedTest
+from unit.simple_model import SimpleModel, random_dataloader
+
+
+def get_config_dict(zero_stage):
+    config_dict = {
+        "train_micro_batch_size_per_gpu": 2,
+        "gradient_accumulation_steps": 1,
+        "steps_per_print": 1,
+        "zero_optimization": {
+            "stage": zero_stage,
+        },
+        "optimizer": {
+            "type": "Adam",
+            "params": {
+                "lr": 1e-3
+            }
+        },
+    }
+
+    if zero_stage == 3:
+        config_dict["zero_optimization"]["stage3_param_persistence_threshold"] = 0
+
+    if get_accelerator().is_bf16_supported():
+        config_dict["bf16"] = {"enabled": True}
+    elif get_accelerator().is_fp16_supported():
+        config_dict["fp16"] = {"enabled": True, "initial_scale_power": 8}
+
+    return config_dict
+
+
+class TestHookCountRegression(DistributedTest):
+    """Test that count_used_parameters_in_backward is not called per-hook."""
+    world_size = 2
+
+    @pytest.mark.parametrize("zero_stage", [2, 3])
+    def test_non_reentrant_single_count_call(self, zero_stage):
+        """Non-reentrant backward should call count_used_parameters_in_backward exactly once."""
+        hidden_dim = 16
+        model = SimpleModel(hidden_dim)
+        config = get_config_dict(zero_stage)
+        engine, _, _, _ = deepspeed.initialize(model=model, model_parameters=model.parameters(), config=config)
+
+        data_loader = random_dataloader(model=engine, total_samples=4, hidden_dim=hidden_dim, device=engine.device)
+
+        # Determine the correct module path to patch based on stage
+        if zero_stage == 2:
+            patch_target = "deepspeed.runtime.zero.stage_1_and_2.count_used_parameters_in_backward"
+        else:
+            patch_target = "deepspeed.runtime.zero.stage3.count_used_parameters_in_backward"
+
+        call_counts = []
+
+        for batch in data_loader:
+            with patch(patch_target, wraps=deepspeed.runtime.utils.count_used_parameters_in_backward) as mock_count:
+                loss = engine(batch[0], batch[1])
+                engine.backward(loss)
+                call_counts.append(mock_count.call_count)
+            engine.step()
+            break
+
+        # Non-reentrant: exactly 1 call per backward
+        assert call_counts[0] == 1, (f"Expected exactly 1 call to count_used_parameters_in_backward "
+                                     f"per backward, got {call_counts[0]}")
+
+    @pytest.mark.parametrize("zero_stage", [2, 3])
+    def test_training_step_succeeds_after_fix(self, zero_stage):
+        """Verify a full training step produces a finite loss after the caching fix."""
+        hidden_dim = 16
+        model = SimpleModel(hidden_dim)
+        config = get_config_dict(zero_stage)
+        engine, _, _, _ = deepspeed.initialize(model=model, model_parameters=model.parameters(), config=config)
+
+        data_loader = random_dataloader(model=engine, total_samples=8, hidden_dim=hidden_dim, device=engine.device)
+
+        losses = []
+        for i, batch in enumerate(data_loader):
+            loss = engine(batch[0], batch[1])
+            assert torch.isfinite(loss), f"Loss is not finite at step {i}: {loss.item()}"
+            losses.append(loss.item())
+            engine.backward(loss)
+            engine.step()
+            if i >= 1:
+                break
+
+        assert len(losses) >= 2, "Expected at least 2 training steps"