[EPLB] Optimize EPLB with numpy (#29499)

Signed-off-by: ilmarkov <markovilya197@gmail.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

tests/distributed/test_eplb_algo.py

@@ -310,3 +310,143 @@ if __name__ == "__main__":
     print(phy2log)
 
     test_basic_rebalance()
+
+
+def _make_phy_replicas_idx_from_phy2log(phy2log: torch.Tensor) -> torch.Tensor:
+    """Create replicas indices mapping from phy2log"""
+    pr = torch.zeros_like(phy2log)
+    for layer in range(phy2log.shape[0]):
+        seen: dict[int, int] = {}
+        row = phy2log[layer].tolist()
+        for i, expert in enumerate(row):
+            r = seen.get(expert, 0)
+            pr[layer, i] = r
+            seen[expert] = r + 1
+    return pr
+
+
+def _validate_intragpu_rearrangement(
+    old_global_expert_indices: torch.Tensor,
+    new_phy2log: torch.Tensor,
+    new_phy_replicas_idx: torch.Tensor,
+    post_phy2log: torch.Tensor,
+    post_phy_replicas_idx: torch.Tensor,
+    num_ranks: int,
+    slots_per_gpu: int,
+):
+    # Per-GPU checks
+    for gpu_idx in range(num_ranks):
+        start = gpu_idx * slots_per_gpu
+        end = start + slots_per_gpu
+        old_seg = old_global_expert_indices[0, start:end]
+        new_seg = new_phy2log[0, start:end]
+        new_rnk = new_phy_replicas_idx[0, start:end]
+        post_seg = post_phy2log[0, start:end]
+        post_rnk = post_phy_replicas_idx[0, start:end]
+
+        # Pairwise equality for (expert, rank) pairs to ensure nothing is lost
+        def sorted_pairs(seg: torch.Tensor, rnk: torch.Tensor):
+            pairs = list(zip(seg.tolist(), rnk.tolist()))
+            pairs.sort()
+            return pairs
+
+        assert sorted_pairs(post_seg, post_rnk) == sorted_pairs(new_seg, new_rnk), (
+            f"Per-GPU pairs of (expert,rank) must match new mapping for GPU {gpu_idx}"
+        )
+
+        # For experts that remain on the same GPU, the old slot is preserved
+        # for at least one occurrence; rank at that slot must be valid for that expert
+        old_list = old_seg.tolist()
+        new_list = new_seg.tolist()
+        post_list = post_seg.tolist()
+        remained = set(old_list) & set(new_list)
+        new_ranks_for_expert: dict[int, list[int]] = {}
+        for v, r in zip(new_list, new_rnk.tolist()):
+            new_ranks_for_expert.setdefault(v, []).append(r)
+        for expert in remained:
+            old_pos = old_list.index(expert)
+            assert post_list[old_pos] == expert, (
+                f"Expert {expert} on GPU {gpu_idx} should stay at old slot {old_pos}"
+            )
+            # Rank at preserved slot must be one of the ranks
+            # the expert has in new mapping
+            assert post_rnk.tolist()[old_pos] in new_ranks_for_expert[expert], (
+                f"Rank for expert {expert} at preserved slot on GPU {gpu_idx} "
+                "must come from new mapping"
+            )
+
+
+@pytest.mark.parametrize(
+    "num_ranks, slots_per_gpu, old_phy2log, new_phy2log",
+    [
+        pytest.param(
+            # Setup: 2 GPUs, 4 slots each, 1 layer
+            # Old mapping: GPU0 -> [0,1,2,3], GPU1 -> [4,5,6,7]
+            # New mapping shuffles within GPU0 and brings 4,5 into GPU0.
+            # GPU0 new -> [1,5,0,4]; GPU1 new -> [6,2,7,3]
+            2,
+            4,
+            torch.tensor([[0, 1, 2, 3, 4, 5, 6, 7]]),
+            torch.tensor([[1, 5, 0, 4, 6, 2, 7, 3]]),
+            id="simple",
+        ),
+        pytest.param(
+            # Setup: 2 GPUs, 5 slots each (total 10 physical experts), 1 layer
+            # Old mapping:
+            #   GPU0 -> [0, 1, 0, 2, 3]  (expert 0 duplicated)
+            #   GPU1 -> [4, 5, 6, 1, 2]
+            # New mapping reorders within GPUs and moves some experts across GPUs,
+            # while still including duplicates:
+            #   GPU0 new -> [0, 5, 4, 0, 1]  (expert 0 duplicated, 4/5 incoming)
+            #   GPU1 new -> [6, 2, 3, 2, 1]  (expert 2 duplicated)
+            2,
+            5,
+            torch.tensor([[0, 1, 0, 2, 3, 4, 5, 6, 1, 2]]),
+            torch.tensor([[0, 5, 4, 0, 1, 6, 2, 3, 2, 1]]),
+            id="duplicates",
+        ),
+        pytest.param(
+            # Setup: 3 GPUs, 4 slots each (total 12 physical experts), 1 layer
+            # Old mapping:
+            #   GPU0 -> [0, 1, 2, 3]
+            #   GPU1 -> [0, 1, 2, 3]
+            #   GPU2 -> [0, 1, 2, 3]
+            # New mapping decides to use one expert on 2 GPUs and shuffles
+            # experts on the third GPU,
+            #   GPU0 new -> [0, 0, 0, 0]
+            #   GPU1 new -> [0, 0, 0, 0]
+            #   GPU2 new -> [1, 2, 3, 0]
+            3,
+            4,
+            torch.tensor([[0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3]]),
+            torch.tensor([[0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 3, 0]]),
+            id="skewed_expert",
+        ),
+    ],
+)
+def test_preserve_intragpu_slots(
+    num_ranks: int,
+    slots_per_gpu: int,
+    old_phy2log: torch.Tensor,
+    new_phy2log: torch.Tensor,
+):
+    """Experts that stay on a GPU keep their old slots; incoming not lost."""
+    phy_replicas_idx = _make_phy_replicas_idx_from_phy2log(new_phy2log)
+
+    post_phy2log, post_phy_replicas_idx = DefaultEplbPolicy.preserve_intragpu_slots(
+        new_phy2log, phy_replicas_idx, num_ranks, old_phy2log
+    )
+
+    # Shapes preserved
+    assert post_phy2log.shape == new_phy2log.shape
+    assert post_phy_replicas_idx.shape == phy_replicas_idx.shape
+
+    _validate_intragpu_rearrangement(
+        old_phy2log,
+        new_phy2log,
+        phy_replicas_idx,
+        post_phy2log,
+        post_phy_replicas_idx,
+        num_ranks,
+        slots_per_gpu,
+    )

tests/distributed/test_eplb_execute.py

@@ -286,15 +286,17 @@ def _test_async_transfer_layer_without_mtp_worker(
         device,
         old_indices,
     )
+    old_indices_cpu = old_indices.cpu()
+    new_indices_cpu = new_indices.cpu()
 
     expert_buffer = [torch.empty_like(w) for w in expert_weights[0]]
     cuda_stream = torch.cuda.Stream(device=device)
 
     for layer_idx in range(num_layers):
-        is_unchanged, is_received_locally, experts_recv_loc = asyncio.run(
+        is_unchanged, is_received_locally, recv_metadata = asyncio.run(
             transfer_layer(
-                old_global_expert_indices=old_indices,
+                old_global_expert_indices=old_indices_cpu,
-                new_global_expert_indices=new_indices,
+                new_global_expert_indices=new_indices_cpu,
                 expert_weights=expert_weights,
                 expert_weights_buffer=expert_buffer,
                 ep_group=ep_group,
@@ -302,16 +304,15 @@ def _test_async_transfer_layer_without_mtp_worker(
                 cuda_stream=cuda_stream,
             )
         )
-
         cuda_stream.synchronize()
         move_from_buffer(
             expert_weights=expert_weights[layer_idx],
-            expert_weights_buffer=expert_buffer,
+            expert_weights_buffers=expert_buffer,
             is_unchanged=is_unchanged,
             is_received_locally=is_received_locally,
-            experts_recv_loc=experts_recv_loc,
+            recv_metadata=recv_metadata,
-            new_indices=new_indices[layer_idx].tolist(),
+            new_indices=new_indices_cpu[layer_idx],
-            ep_group=ep_group,
+            ep_rank=ep_rank,
         )
 
     verify_expert_weights_after_shuffle(

vllm/config/parallel.py

@@ -69,6 +69,10 @@ class EPLBConfig:
     Log the balancedness each step of expert parallelism.
     This is turned off by default since it will cause communication overhead.
     """
+    log_balancedness_interval: int = 1
+    """
+    Interval for logging the balancedness.
+    """
     use_async: bool = False
     """
     Whether to use non-blocking EPLB.
@@ -77,6 +81,14 @@ class EPLBConfig:
     policy: EPLBPolicyOption = "default"
     """The policy type for expert parallel load balancing (EPLB)."""
 
+    @model_validator(mode="after")
+    def _validate_eplb_config(self) -> Self:
+        if self.use_async and self.policy != "default":
+            raise ValueError("Async EPLB is only supported with the default policy.")
+        if self.log_balancedness and self.log_balancedness_interval <= 0:
+            raise ValueError("log_balancedness_interval must be greater than 0.")
+        return self
+
 
 @config
 @dataclass

vllm/distributed/eplb/async_worker.py

@@ -89,7 +89,7 @@ async def transfer_run_periodically(
                         (
                             model_state.is_unchanged,
                             model_state.is_received_locally,
-                            model_state.experts_recv_loc,
+                            model_state.recv_metadata,
                         ) = await transfer_layer(
                             old_global_expert_indices=model_state.physical_to_logical_map,
                             new_global_expert_indices=model_state.new_physical_to_logical_map,

vllm/distributed/eplb/eplb_state.py

@@ -27,10 +27,10 @@ physical experts.
 """
 
 import threading
-import time
 from collections.abc import Sequence
 from dataclasses import dataclass
 
+import numpy as np
 import torch
 from torch.distributed import ProcessGroup, all_reduce
 
@@ -46,7 +46,11 @@ from vllm.model_executor.models.interfaces import MixtureOfExperts
 
 from .async_worker import start_async_worker
 from .policy import EPLB_POLICIES, AbstractEplbPolicy, DefaultEplbPolicy
-from .rebalance_execute import move_from_buffer, rearrange_expert_weights_inplace
+from .rebalance_execute import (
+    RecvMetadata,
+    move_from_buffer,
+    rearrange_expert_weights_inplace,
+)
 
 logger = init_logger(__name__)
 
@@ -164,20 +168,19 @@ class EplbModelState:
     """
     Whether the async EPLB needs to poll peers for buffer readiness.
     """
-    is_unchanged: list[bool]
+    is_unchanged: np.ndarray
     """
     intermediate variable between `move_to_buffer` and `move_to_workspace`.
     The size is same as the num of physical experts in the current layer.
     """
-    is_received_locally: list[bool]
+    is_received_locally: np.ndarray
     """
     intermediate variable between `move_to_buffer` and `move_to_workspace`.
     The size is same as the num of physical experts in the current layer.
     """
-    experts_recv_loc: dict[int, int]
+    recv_metadata: RecvMetadata
     """
     intermediate variable between `move_to_buffer` and `move_to_workspace`.
-    The size is same as the num of physical experts in the current layer.
     """
     is_async_enabled: bool
     """
@@ -507,9 +510,14 @@ class EplbState:
             layer_to_transfer=0,
             rebalanced=False,
             pending_global_ready_check=False,
-            is_unchanged=[],
+            is_unchanged=np.array([]),
-            is_received_locally=[],
+            is_received_locally=np.array([]),
-            experts_recv_loc={},
+            recv_metadata=RecvMetadata(
+                recv_primary_mask=np.array([]),
+                recv_count=0,
+                recv_expert_ids=np.array([]),
+                recv_dst_rows=np.array([]),
+            ),
             is_async_enabled=self.is_async,
             cuda_device_index=self.cuda_device_index,
             new_physical_to_logical_map=new_physical_to_logical_map,
@@ -553,7 +561,12 @@ class EplbState:
             for eplb_model_state in self.model_states.values():
                 eplb_model_state.expert_load_pass.zero_()
 
-        if log_stats:
+        if (
+            log_stats
+            and self.expert_rearrangement_step
+            % self.parallel_config.eplb_config.log_balancedness_interval
+            == 0
+        ):
             # Sync the expert load pass for each model (main and drafter).
             # expert_load_pass: (num_moe_layers, num_physical_experts)
             expert_load_pass_list = self._sync_load_pass()
@@ -586,12 +599,15 @@ class EplbState:
                 if ep_group.rank() == 0:
                     logger.info(
                         "EPLB step: %d for model %s: avg_tokens=%.2f, "
-                        "max_tokens=%d, balancedness=%.4f",
+                        "max_tokens=%d, balancedness=%.4f, "
+                        "steps until the next rearrangement: %d",
                         self.expert_rearrangement_step,
                         eplb_model_state.model_name,
                         avg_tokens,
                         max_tokens,
                         balancedness,
+                        self.expert_rearrangement_step_interval
+                        - self.expert_rearrangement_step,
                     )
 
         # Update the expert load sliding window
@@ -684,11 +700,14 @@ class EplbState:
         ep_group = get_ep_group().device_group
         ep_rank = ep_group.rank()
 
-        time_start = None
+        start_event = None
+        end_event = None
         is_main_rank = ep_rank == 0
         if is_main_rank:
-            torch.cuda.synchronize()
+            if not self.is_async or is_profile:
-            time_start = time.perf_counter()
+                start_event = torch.cuda.Event(enable_timing=True)
+                end_event = torch.cuda.Event(enable_timing=True)
+                start_event.record()
             logger.info(
                 "Rearranging experts %s %s...",
                 "(async mode)" if self.is_async else "sync mode",
@@ -800,6 +819,7 @@ class EplbState:
                 num_groups,
                 num_nodes,
                 num_gpus,
+                eplb_model_state.physical_to_logical_map,
             )
 
             if not eplb_model_state.is_async_enabled or is_profile:
@@ -848,17 +868,17 @@ class EplbState:
                         new_logical_replica_count
                     )
                 if is_main_rank:
-                    assert time_start is not None
+                    assert start_event is not None
-                    torch.cuda.synchronize()
+                    assert end_event is not None
-                    time_end = time.perf_counter()
+                    end_event.record()
+                    end_event.synchronize()
+                    gpu_elapsed = start_event.elapsed_time(end_event) / 1000.0
                     logger.info(
-                        "Rearranged experts%sin %.2f seconds.",
+                        "Rearranged experts %s in %.2f s.",
                         " (profile) " if is_profile else " ",
-                        time_end - time_start,
+                        gpu_elapsed,
                     )
             else:
-                device = eplb_model_state.physical_to_logical_map.device
-                new_physical = new_physical_to_logical_map.to(device)
                 max_slots = eplb_model_state.logical_to_physical_map.shape[-1]
                 padded_logical = torch.nn.functional.pad(
                     new_logical_to_physical_map,
@@ -869,7 +889,10 @@ class EplbState:
                     eplb_model_state.logical_replica_count.device
                 )
 
-                eplb_model_state.new_physical_to_logical_map = new_physical
+                # Move map to cpu in advance
+                eplb_model_state.new_physical_to_logical_map = (
+                    new_physical_to_logical_map.cpu()
+                )
                 eplb_model_state.new_logical_to_physical_map = padded_logical
                 eplb_model_state.new_logical_replica_count = new_replica
 
@@ -968,25 +991,30 @@ class EplbState:
                 stream = torch.cuda.current_stream(device=device_index)
                 stream.wait_event(model_state.buffer_ready_event)
                 model_state.buffer_ready_event = None
+            expert_weights = model_state.model.expert_weights[
+                model_state.layer_to_transfer
+            ]
+            expert_weights_buffer = model_state.expert_buffer
+            new_indices = (
+                model_state.new_physical_to_logical_map[model_state.layer_to_transfer]
+                .cpu()
+                .numpy()
+            )
             move_from_buffer(
-                expert_weights=model_state.model.expert_weights[
+                expert_weights=expert_weights,
-                    model_state.layer_to_transfer
+                expert_weights_buffers=expert_weights_buffer,
-                ],
-                expert_weights_buffer=model_state.expert_buffer,
                 is_unchanged=model_state.is_unchanged,
                 is_received_locally=model_state.is_received_locally,
-                experts_recv_loc=model_state.experts_recv_loc,
+                recv_metadata=model_state.recv_metadata,
-                new_indices=model_state.new_physical_to_logical_map[
+                new_indices=new_indices,
-                    model_state.layer_to_transfer
+                ep_rank=ep_group.rank(),
-                ].tolist(),
-                ep_group=ep_group,
             )
             transferred_layer = model_state.layer_to_transfer
             self._update_layer_mapping_from_new(model_state, transferred_layer)
             # After the main thread consumes, advance layer_to_transfer
             model_state.layer_to_transfer += 1
             model_state.ep_buffer_ready = 0
-            logger.info(
+            logger.debug(
                 "model %s successfully move_to_workspace layer %d",
                 model_state.model_name,
                 transferred_layer,

vllm/distributed/eplb/policy/abstract.py

@@ -16,6 +16,7 @@ class AbstractEplbPolicy(ABC):
         num_groups: int,
         num_nodes: int,
         num_ranks: int,
+        old_global_expert_indices: torch.Tensor | None = None,
     ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
         """
         Entry point for expert-parallelism load balancer.
@@ -28,7 +29,9 @@ class AbstractEplbPolicy(ABC):
             num_groups: number of expert groups
             num_nodes: number of server nodes
             num_ranks: number of ranks, must be a multiple of `num_nodes`
-
+            old_global_expert_indices: [layers, num_logical_experts], the old global
+                expert indices. Used to avoid unnecessary weight copying
+                for experts moving within one rank.
         Returns:
             physical_to_logical_map: [layers, num_replicas], the expert
                 index of each replica

vllm/distributed/eplb/policy/default.py

@@ -93,7 +93,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
 
         Returns:
             phy2log: [X, num_phy], logical expert id of each physical expert
-            rank: [X, num_phy], the replica rank
+            replica_idx: [X, num_phy], the index of the replica for each logical expert
             logcnt: [X, num_log], number of replicas for each logical expert
         """
         n, num_log = weight.shape
@@ -101,15 +101,15 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         assert num_redundant >= 0
         device = weight.device
         phy2log = torch.arange(num_phy, dtype=torch.int64, device=device).repeat(n, 1)
-        rank = torch.zeros(n, num_phy, dtype=torch.int64, device=device)
+        replica_idx = torch.zeros(n, num_phy, dtype=torch.int64, device=device)
         logcnt = torch.ones(n, num_log, dtype=torch.int64, device=device)
         arangen = torch.arange(n, dtype=torch.int64, device=device)
         for i in range(num_log, num_phy):
             redundant_indices = (weight / logcnt).max(dim=-1).indices
             phy2log[:, i] = redundant_indices
-            rank[:, i] = logcnt[arangen, redundant_indices]
+            replica_idx[:, i] = logcnt[arangen, redundant_indices]
             logcnt[arangen, redundant_indices] += 1
-        return phy2log, rank, logcnt
+        return phy2log, replica_idx, logcnt
 
     @classmethod
     def rebalance_experts_hierarchical(
@@ -132,7 +132,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         Returns:
             phy2log: [layers, num_replicas], the expert
                 index of each replica
-            log2phy: [layers, num_logical_experts, X],
+            pphy_replicas_idx: [layers, num_logical_experts, X],
                 the replica indices for each expert
             logcnt: [layers, num_logical_experts], number of
                 physical replicas for each logical expert
@@ -177,7 +177,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         tokens_per_mlog = weight.gather(-1, mlog2log).view(
             -1, num_logical_experts // num_nodes
         )
-        phy2mlog, phyrank, mlogcnt = cls.replicate_experts(
+        phy2mlog, replicas_idx, mlogcnt = cls.replicate_experts(
             tokens_per_mlog, num_physical_experts // num_nodes
         )
 
@@ -203,9 +203,109 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
             ).view(1, -1, 1)
         ).flatten(-2)
         pphy2log = mlog2log.gather(-1, pphy2mlog)
-        pphyrank = phyrank.gather(-1, pphy2phy).view(num_layers, -1)
+        pphy_replicas_idx = replicas_idx.gather(-1, pphy2phy).view(num_layers, -1)
         logcnt = mlogcnt.view(num_layers, -1).gather(-1, log2mlog)
-        return pphy2log, pphyrank, logcnt
+        return pphy2log, pphy_replicas_idx, logcnt
+
+    @classmethod
+    def preserve_intragpu_slots(
+        cls,
+        phy2log: torch.Tensor,
+        phy_replicas_idx: torch.Tensor,
+        num_ranks: int,
+        old_global_expert_indices: torch.Tensor,
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        """
+        Reorder the new mapping per GPU so that experts that remain on the same GPU
+        keep their previous slot positions when possible. Incoming experts to that GPU
+        fill any remaining available slots. This is applied only when the number of GPUs
+        is unchanged and the slots per GPU remain the same between
+        the old and new mappings.
+        """
+        device = phy2log.device
+        num_phy_experts = phy2log.shape[1]
+        if num_ranks <= 0 or num_phy_experts % num_ranks != 0:
+            return phy2log, phy_replicas_idx
+
+        # Move to CPU and convert to NumPy for processing
+        new_phy2log_np = phy2log.cpu().numpy()
+        replicas_idx_np = phy_replicas_idx.cpu().numpy()
+        old_phy2log_np = old_global_expert_indices.cpu().numpy()
+
+        slots_per_gpu = num_phy_experts // num_ranks
+        num_layers = new_phy2log_np.shape[0]
+
+        post_phy2log_np = new_phy2log_np.copy()
+        post_phy_replicas_idx_np = replicas_idx_np.copy()
+
+        for gpu_idx in range(num_ranks):
+            start = gpu_idx * slots_per_gpu
+            end = start + slots_per_gpu
+            # Experts across all layers for this GPU
+            old_local = old_phy2log_np[:, start:end]  # [layers, slots]
+            new_local = new_phy2log_np[:, start:end]  # [layers, slots]
+            new_ridx = replicas_idx_np[:, start:end]  # [layers, slots]
+
+            used_new_indices = np.zeros((num_layers, slots_per_gpu), dtype=bool)
+            preserved_positions = np.zeros((num_layers, slots_per_gpu), dtype=bool)
+
+            # First pass: preserve same-logical experts in their previous slots
+            for slot_idx in range(slots_per_gpu):
+                # matches: [layers, slots], True where new local experts have
+                # the same logical value as the old from 'slot_idx' and not checked yet
+                matches = (new_local == old_local[:, slot_idx][:, None]) & (
+                    ~used_new_indices
+                )
+                has_any = matches.any(axis=1)
+                if np.any(has_any):
+                    first_idx = np.argmax(matches, axis=1)
+                    layer_indices = np.nonzero(has_any)[0]
+                    matched_new_positions = first_idx[layer_indices]
+                    post_phy2log_np[layer_indices, start + slot_idx] = new_local[
+                        layer_indices, matched_new_positions
+                    ]
+                    post_phy_replicas_idx_np[layer_indices, start + slot_idx] = (
+                        new_ridx[layer_indices, matched_new_positions]
+                    )
+                    used_new_indices[layer_indices, matched_new_positions] = True
+                    preserved_positions[layer_indices, slot_idx] = True
+
+            # Second pass: fill remaining slots with remaining new experts
+            remaining_mask = ~used_new_indices  # [layers, slots]
+            fill_mask = ~preserved_positions  # [layers, slots]
+            if remaining_mask.any() and fill_mask.any():
+                idx_base = np.tile(np.arange(slots_per_gpu), (num_layers, 1))
+                # Sentinel value for unavailable positions.
+                large = slots_per_gpu + 1
+                # Priorities: keep original index for available spots, set sentinel
+                # for unavailable; lower is earlier.
+                remaining_priority = np.where(remaining_mask, idx_base, large)
+                fill_priority = np.where(fill_mask, idx_base, large)
+                # Sort to get ordered indices of available src/dst positions per layer.
+                remaining_indices = np.argsort(remaining_priority, axis=1)
+                fill_indices = np.argsort(fill_priority, axis=1)
+                # Fill count per layer (cannot exceed either side).
+                remaining_counts = remaining_mask.sum(axis=1)
+                fill_counts = fill_mask.sum(axis=1)
+                take_counts = np.minimum(remaining_counts, fill_counts)
+                # Assign remaining new experts to remaining slots per layer.
+                for layer_idx in range(num_layers):
+                    k = int(take_counts[layer_idx])
+                    if k <= 0:
+                        continue
+                    src_pos = remaining_indices[layer_idx, :k]
+                    dst_pos = fill_indices[layer_idx, :k]
+                    post_phy2log_np[layer_idx, start + dst_pos] = new_local[
+                        layer_idx, src_pos
+                    ]
+                    post_phy_replicas_idx_np[layer_idx, start + dst_pos] = new_ridx[
+                        layer_idx, src_pos
+                    ]
+
+        # Convert back to torch and move to original device
+        post_phy2log = torch.from_numpy(post_phy2log_np).to(device)
+        post_phy_replicas_idx = torch.from_numpy(post_phy_replicas_idx_np).to(device)
+        return post_phy2log, post_phy_replicas_idx
 
     @classmethod
     def rebalance_experts(
@@ -215,6 +315,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         num_groups: int,
         num_nodes: int,
         num_ranks: int,
+        old_global_expert_indices: torch.Tensor | None = None,
     ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
         """
         Entry point for expert-parallelism load balancer.
@@ -228,7 +329,9 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
             num_nodes: number of server nodes, where the intra-node network
                 (e.g, NVLink) is faster
             num_ranks: number of ranks, must be a multiple of `num_nodes`
-
+            old_global_expert_indices: [layers, num_logical_experts], the old global
+                expert indices. Used to avoid unnecessary weight copying
+                for experts moving within one rank.
         Returns:
             phy2log: [layers, num_replicas], the expert
                 index of each replica
@@ -241,14 +344,23 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         weight = weight.float()
         if num_groups % num_nodes == 0:
             # use hierarchical load-balance policy
-            phy2log, phyrank, logcnt = cls.rebalance_experts_hierarchical(
+            phy2log, phy_replicas_idx, logcnt = cls.rebalance_experts_hierarchical(
                 weight, num_replicas, num_groups, num_nodes, num_ranks
             )
         else:
             # use global load-balance policy
-            phy2log, phyrank, logcnt = cls.rebalance_experts_hierarchical(
+            phy2log, phy_replicas_idx, logcnt = cls.rebalance_experts_hierarchical(
                 weight, num_replicas, 1, 1, num_ranks
             )
+        # Optional postprocessing to preserve slots for experts moving
+        # within the same GPU
+        # Only apply when the number of GPUs and slots per GPU remain unchanged.
+        # Helps to avoid unnecessary weight copying when experts move
+        # within the same GPU.
+        if old_global_expert_indices is not None:
+            phy2log, phy_replicas_idx = cls.preserve_intragpu_slots(
+                phy2log, phy_replicas_idx, num_ranks, old_global_expert_indices
+            )
         num_redundant_experts = num_replicas - num_logical_experts
         maxlogcnt = num_redundant_experts + 1
         log2phy: torch.Tensor = torch.full(
@@ -259,7 +371,7 @@ class DefaultEplbPolicy(AbstractEplbPolicy):
         )
         log2phy.view(num_layers, -1).scatter_(
             -1,
-            phy2log * maxlogcnt + phyrank,
+            phy2log * maxlogcnt + phy_replicas_idx,
             torch.arange(num_replicas, dtype=torch.int64, device=log2phy.device).expand(
                 num_layers, -1
             ),

vllm/distributed/eplb/rebalance_execute.py

@@ -6,9 +6,10 @@ The actual execution of the rearrangement.
 This involves the exchange of expert weights between GPUs.
 """
 
-from collections.abc import Iterable, MutableSequence, Sequence
+from collections.abc import Iterable, Sequence
-from functools import partial
+from dataclasses import dataclass
 
+import numpy as np
 import torch
 from torch.distributed import (
     P2POp,
@@ -18,214 +19,318 @@ from torch.distributed import (
     get_global_rank,
 )
 
+from vllm.logger import init_logger
 
-def idx_local_to_global(
+logger = init_logger(__name__)
-    local_idx: int,
-    local_cnt: int,
-    ep_rank: int,
-) -> int:
-    """
-    Convert a local expert index to a global expert index.
-    """
-    return ep_rank * local_cnt + local_idx
 
 
-def idx_global_to_local(
+@dataclass
-    global_idx: int,
+class RecvMetadata:
-    local_cnt: int,
+    """Metadata describing remote receives during EPLB rebalancing."""
-    ep_rank: int,
+
-) -> int:
+    recv_primary_mask: np.ndarray
-    """
+    """Mask of (num_local_experts,) indicating primary experts received."""
-    Convert a global expert index to a local expert index.
+    recv_count: int
-    """
+    """Number of received experts for the layer."""
-    return global_idx - ep_rank * local_cnt
+    recv_expert_ids: np.ndarray
+    """Expert ids (num_local_experts,) of remote primary experts."""
+    recv_dst_rows: np.ndarray
+    """Target expert indices (num_local_experts,) in local tensors to send."""
 
 
-def global_idx_to_rank(
+# Type alias for the result of move_to_buffer or transfer_layer
-    global_idx: int,
+MoveToBufferResult = tuple[np.ndarray, np.ndarray, RecvMetadata]
-    local_cnt: int,
-) -> int:
-    """
-    Convert a global expert index to a rank index.
-    """
-    return global_idx // local_cnt
 
 
-def get_ep_ranks_with_expert(
+def get_ep_ranks_with_experts_batch(
-    idx: int,
+    expert_ids: np.ndarray,
     num_local_experts: int,
-    old_indices: Sequence[int],
+    old_indices: np.ndarray,
-    new_indices: Sequence[int],
+    new_indices: np.ndarray,
-) -> tuple[MutableSequence[int], MutableSequence[int]]:
+) -> tuple[dict[int, list[int]], dict[int, list[int]]]:
     """
     Get the ranks of the experts that need to be exchanged.
 
     Args:
-        idx: The index of the expert.
+        expert_ids: 1D array of expert indices to query.
         num_local_experts: The number of local experts.
         old_indices: The old indices of the experts.
         new_indices: The new indices of the experts.
 
     Returns:
-        A tuple of two lists:
+        A tuple of two dictionaries mapping expert_id to:
-        - The ranks of the experts that need to be sent.
+        - ranks_to_send: The ranks that have this expert and need to send.
-        - The ranks of the experts that need to be received.
+        - ranks_to_recv: The ranks that need to receive this expert.
     """
-    global2rank = partial(
+    ranks_to_send_map: dict[int, list[int]] = {}
-        global_idx_to_rank,
+    ranks_to_recv_map: dict[int, list[int]] = {}
-        local_cnt=num_local_experts,
-    )
 
-    ranks_to_send: list[int] = []
+    # Fast path: if no experts, return empty dicts
-    ranks_to_recv: list[int] = []
+    if expert_ids.size == 0:
+        return ranks_to_send_map, ranks_to_recv_map
 
-    for i, e in enumerate(old_indices):
+    unique_experts = np.unique(expert_ids)
-        if e == idx:
+    num_positions = len(old_indices)
-            rank = global2rank(i)
+    position_indices = np.arange(num_positions, dtype=np.int32)
-            if not ranks_to_send or ranks_to_send[-1] != rank:
-                ranks_to_send.append(rank)
 
-    for i, e in enumerate(new_indices):
+    # Vectorized approach: find all positions matching any query expert in one pass
-        if e == idx:
+    # Use np.isin to get boolean masks for all relevant positions at once
-            rank = global2rank(i)
+    old_relevant_mask = np.isin(old_indices, unique_experts)
-            if not ranks_to_recv or ranks_to_recv[-1] != rank:
+    new_relevant_mask = np.isin(new_indices, unique_experts)
-                ranks_to_recv.append(rank)
 
-    # Remove those ranks that can get this expert locally.
+    # Process old_indices (send ranks)
-    ranks_to_send_set = set(ranks_to_send)
+    if np.any(old_relevant_mask):
-    ranks_to_recv_actual = [
+        old_relevant_positions = position_indices[old_relevant_mask]
-        rank for rank in ranks_to_recv if rank not in ranks_to_send_set
+        old_relevant_experts = old_indices[old_relevant_mask]
-    ]
+        old_relevant_ranks = old_relevant_positions // num_local_experts
 
-    return ranks_to_send, ranks_to_recv_actual
+        # Sort by expert first, then by position (to maintain first-appearance order)
+        sort_order = np.lexsort((old_relevant_positions, old_relevant_experts))
+        sorted_experts = old_relevant_experts[sort_order]
+        sorted_ranks = old_relevant_ranks[sort_order]
+
+        # Find boundaries where expert changes
+        expert_boundaries = np.concatenate(
+            [[0], np.where(np.diff(sorted_experts) != 0)[0] + 1, [len(sorted_experts)]]
+        )
+
+        # For each expert, extract unique ranks in order of first appearance
+        for i in range(len(expert_boundaries) - 1):
+            start, end = expert_boundaries[i], expert_boundaries[i + 1]
+            expert = int(sorted_experts[start])
+            expert_ranks = sorted_ranks[start:end]
+
+            # Get unique ranks preserving order
+            _, unique_idx = np.unique(expert_ranks, return_index=True)
+            unique_ranks = expert_ranks[np.sort(unique_idx)]
+            ranks_to_send_map[expert] = unique_ranks.tolist()
+
+    # Process new_indices (recv ranks)
+    if np.any(new_relevant_mask):
+        new_relevant_positions = position_indices[new_relevant_mask]
+        new_relevant_experts = new_indices[new_relevant_mask]
+        new_relevant_ranks = new_relevant_positions // num_local_experts
+
+        # Sort by expert first, then by position
+        sort_order = np.lexsort((new_relevant_positions, new_relevant_experts))
+        sorted_experts = new_relevant_experts[sort_order]
+        sorted_ranks = new_relevant_ranks[sort_order]
+
+        # Find boundaries where expert changes
+        expert_boundaries = np.concatenate(
+            [[0], np.where(np.diff(sorted_experts) != 0)[0] + 1, [len(sorted_experts)]]
+        )
+
+        # For each expert, extract unique ranks and exclude local copies
+        for i in range(len(expert_boundaries) - 1):
+            start, end = expert_boundaries[i], expert_boundaries[i + 1]
+            expert = int(sorted_experts[start])
+            expert_ranks = sorted_ranks[start:end]
+
+            # Get unique ranks preserving order
+            _, unique_idx = np.unique(expert_ranks, return_index=True)
+            unique_ranks = expert_ranks[np.sort(unique_idx)]
+
+            # Remove ranks that have local copies (in send map)
+            send_ranks_set = set(ranks_to_send_map.get(expert, []))
+            recv_ranks_actual = [
+                int(r) for r in unique_ranks if r not in send_ranks_set
+            ]
+            ranks_to_recv_map[expert] = recv_ranks_actual
+
+    # Handle experts that only appear in old (send only) or new (recv only)
+    for expert in unique_experts:
+        expert = int(expert)
+        if expert not in ranks_to_send_map:
+            ranks_to_send_map[expert] = []
+        if expert not in ranks_to_recv_map:
+            ranks_to_recv_map[expert] = []
+
+    return ranks_to_send_map, ranks_to_recv_map
 
 
 def move_to_buffer(
     num_local_experts: int,
-    old_indices: Sequence[int],
+    old_indices: np.ndarray,
-    new_indices: Sequence[int],
+    new_indices: np.ndarray,
     expert_weights: Iterable[torch.Tensor],
-    expert_weights_buffer: Sequence[torch.Tensor],
+    expert_weights_buffers: Sequence[torch.Tensor],
     cuda_stream: torch.cuda.Stream | None,
     ep_group: ProcessGroup,
-) -> tuple[list[bool], list[bool], dict[int, int]]:
+) -> MoveToBufferResult:
     """
-    Perform expert weights rearrangement of one layer.
+    Rearranges expert weights during EPLB rebalancing.
+
+    Args:
+        num_local_experts: Number of local experts.
+        old_indices: (num_experts_total,) ndarray of current (old)
+            global-to-local expert assignments.
+        new_indices: (num_experts_total,) ndarray of desired (new)
+            global-to-local assignments after rebalance.
+        expert_weights: Original expert weights for the layer.
+        expert_weights_buffers: Intermediate buffers (one per tensor).
+        cuda_stream: CUDA stream for async copies (can be None for sync mode).
+        ep_group: Distributed process group for expert parallel comms.
+
+    Returns:
+        is_unchanged (np.ndarray): (num_local_experts,), True where an expert row
+            is unchanged after rebalance.
+        is_received_locally (np.ndarray): (num_local_experts,), True where a row
+            can be updated from local data.
+        RecvMetadata: Metadata needed for completing remote weight transfers.
     """
+    assert old_indices.shape == new_indices.shape
     ep_rank = ep_group.rank()
-    local2global = partial(
+
-        idx_local_to_global,
+    recv_primary_mask = np.zeros((num_local_experts,), dtype=np.bool_)
-        local_cnt=num_local_experts,
+    send_expert_ids = np.full((num_local_experts,), -1, dtype=np.int64)
-        ep_rank=ep_rank,
+    send_src_rows = np.full((num_local_experts,), -1, dtype=np.int32)
+    recv_expert_ids = np.full((num_local_experts,), -1, dtype=np.int64)
+    recv_dst_rows = np.full((num_local_experts,), -1, dtype=np.int32)
+
+    base = ep_rank * num_local_experts
+    local_rows = np.arange(num_local_experts, dtype=np.int32)
+    local_global = base + local_rows
+
+    old_local_expert_ids = old_indices[local_global]
+    new_local_expert_ids = new_indices[local_global]
+
+    # Unchanged mask
+    is_unchanged = old_local_expert_ids == new_local_expert_ids
+
+    # Local receive eligibility
+    new_valid = new_local_expert_ids != -1
+    can_recv_local = np.isin(
+        new_local_expert_ids, old_local_expert_ids, assume_unique=False
+    )
+    is_received_locally = np.logical_or(
+        is_unchanged, np.logical_and(new_valid, can_recv_local)
     )
 
-    # 0. Do nothing for experts that did not change.
+    # Send map: first src row per unique expert present locally in old mapping
-    is_unchanged = [
+    send_count = 0
-        old_indices[local2global(i)] == new_indices[local2global(i)]
+    valid_old = old_local_expert_ids != -1
-        for i in range(num_local_experts)
+    if np.any(valid_old):
-    ]
+        uniq_experts, first_idx = np.unique(
+            old_local_expert_ids[valid_old], return_index=True
+        )
+        filtered_rows = local_rows[valid_old]
+        src_rows = filtered_rows[first_idx]
+        send_count = int(uniq_experts.shape[0])
+        send_expert_ids[:send_count] = uniq_experts
+        send_src_rows[:send_count] = src_rows
 
-    # 1. Perform weight copy inside the local rank.
+    # Recv map: primary dst per unique expert needed remotely
-    is_received_locally = is_unchanged[:]
+    recv_count = 0
-    for src in range(num_local_experts):
+    need_recv_mask = np.logical_and(~is_received_locally, new_valid)
-        src_global = local2global(src)
+    if np.any(need_recv_mask):
-        for dst in range(num_local_experts):
+        desired_experts = new_local_expert_ids[need_recv_mask]
-            dst_global = local2global(dst)
+        desired_dsts = local_rows[need_recv_mask]
-            if is_received_locally[dst]:
+        uniq_recv_experts, uniq_indices = np.unique(desired_experts, return_index=True)
-                continue
+        dst_rows = desired_dsts[uniq_indices]
-            if old_indices[src_global] == -1 or new_indices[dst_global] == -1:
+        recv_count = int(uniq_recv_experts.shape[0])
-                continue
+        recv_expert_ids[:recv_count] = uniq_recv_experts
-            if old_indices[src_global] == new_indices[dst_global]:
+        recv_dst_rows[:recv_count] = dst_rows
-                is_received_locally[dst] = True
+        recv_primary_mask[dst_rows] = True
-                for weight, buffer in zip(expert_weights, expert_weights_buffer):
+
-                    with torch.cuda.stream(cuda_stream):
+    eligible_local_buffer_mask = np.logical_and(~is_unchanged, is_received_locally)
-                        buffer[dst].copy_(weight[src], non_blocking=True)
+
+    # 1. Local moves into tmp buffers
+    if bool(eligible_local_buffer_mask.any()) and send_count > 0:
+        dest_indices = np.nonzero(eligible_local_buffer_mask)[0].tolist()
+        expert_to_src_map = dict(
+            zip(send_expert_ids[:send_count], send_src_rows[:send_count])
+        )
+        for dst in dest_indices:
+            expert = new_local_expert_ids[dst]
+            src_local = expert_to_src_map.get(expert, -1)
+            if src_local != -1:
+                for w, b in zip(expert_weights, expert_weights_buffers):
+                    b[dst].copy_(w[src_local], non_blocking=True)
 
     p2p_ops: list[P2POp] = []
 
-    # 2. Initiate sending of weights.
+    # Pre-compute global ranks mapping
-    experts_send_loc: dict[int, int] = {}
+    ep_size = ep_group.size()
-    for src in range(num_local_experts):
+    rank_to_global = {rank: get_global_rank(ep_group, rank) for rank in range(ep_size)}
-        expert = old_indices[local2global(src)]
-        if expert == -1:
-            continue
-        if expert in experts_send_loc:
-            continue
-        experts_send_loc[expert] = src
 
-    # We need to sort here to match send/recv
+    # 2. Post sends
-    for expert, src in sorted(experts_send_loc.items()):
+    if send_count > 0:
-        ranks_to_send, ranks_to_recv = get_ep_ranks_with_expert(
+        experts = send_expert_ids[:send_count]
-            expert,
+        srcs = send_src_rows[:send_count]
+        order = np.argsort(experts, kind="stable")
+        experts = experts[order]
+        srcs = srcs[order]
+
+        send_map, recv_map = get_ep_ranks_with_experts_batch(
+            experts,
             num_local_experts,
             old_indices,
             new_indices,
         )
 
-        # Calculate the ranks to send by this rank
+        for expert, src in zip(experts.tolist(), srcs.tolist()):
-        num_dst_per_sender = len(ranks_to_recv) // len(ranks_to_send)
+            ranks_to_send = send_map[expert]
-        sender_pos = ranks_to_send.index(ep_rank)
+            ranks_to_recv = recv_map[expert]
-        recv_begin = sender_pos * num_dst_per_sender
+            if not ranks_to_send or not ranks_to_recv:
-        recv_end = recv_begin + num_dst_per_sender
+                continue
-        recv_ranks = ranks_to_recv[recv_begin:recv_end]
+            num_dst_per_sender = len(ranks_to_recv) // len(ranks_to_send)
+            sender_pos = ranks_to_send.index(ep_rank)
+            recv_begin = sender_pos * num_dst_per_sender
+            recv_end = recv_begin + num_dst_per_sender
+            recv_ranks = ranks_to_recv[recv_begin:recv_end]
+            remainder_start = len(ranks_to_send) * num_dst_per_sender
+            recver_pos = remainder_start + sender_pos
+            if recver_pos < len(ranks_to_recv):
+                recv_ranks.append(ranks_to_recv[recver_pos])
+            for dst in recv_ranks:
+                dst_global = rank_to_global[dst]
+                p2p_ops += [
+                    P2POp(
+                        torch.distributed.isend,
+                        w[src],
+                        dst_global,
+                    )
+                    for w in expert_weights
+                ]
 
-        # Tackle remainders
+    # 3. Post recvs
-        remainder_start = len(ranks_to_send) * num_dst_per_sender
+    if recv_count > 0:
-        recver_pos = remainder_start + sender_pos
+        experts = recv_expert_ids[:recv_count]
-        if recver_pos < len(ranks_to_recv):
+        dsts = recv_dst_rows[:recv_count]
-            recv_ranks.append(ranks_to_recv[recver_pos])
+        order = np.argsort(experts, kind="stable")
+        experts = experts[order]
+        dsts = dsts[order]
 
-        for dst in recv_ranks:
+        send_map, recv_map = get_ep_ranks_with_experts_batch(
-            dst_global = get_global_rank(ep_group, dst)
+            experts,
+            num_local_experts,
+            old_indices,
+            new_indices,
+        )
+
+        for expert, dst in zip(experts.tolist(), dsts.tolist()):
+            ranks_to_send = send_map[expert]
+            ranks_to_recv = recv_map[expert]
+            if not ranks_to_send or not ranks_to_recv:
+                continue
+            num_dst_per_sender = len(ranks_to_recv) // len(ranks_to_send)
+            recver_pos = ranks_to_recv.index(ep_rank)
+            remainder_start = len(ranks_to_send) * num_dst_per_sender
+            if recver_pos < remainder_start:
+                src = ranks_to_send[recver_pos // num_dst_per_sender]
+            else:
+                src = ranks_to_send[recver_pos - remainder_start]
+            src_global = rank_to_global[src]
             p2p_ops += [
                 P2POp(
-                    torch.distributed.isend,
+                    torch.distributed.irecv,
-                    weight[src],
+                    b[dst],
-                    dst_global,
+                    src_global,
                 )
-                for weight in expert_weights
+                for b in expert_weights_buffers
             ]
 
-    # 3. Initiate receiving of weights.
-    experts_recv_loc: dict[int, int] = {}
-    for dst in range(num_local_experts):
-        if is_received_locally[dst]:
-            continue
-        expert = new_indices[local2global(dst)]
-        if expert == -1:
-            continue
-        if expert in experts_recv_loc:
-            continue
-        experts_recv_loc[expert] = dst
-
-    # We need to sort here to match send/recv
-    for expert, dst in sorted(experts_recv_loc.items()):
-        ranks_to_send, ranks_to_recv = get_ep_ranks_with_expert(
-            expert,
-            num_local_experts,
-            old_indices,
-            new_indices,
-        )
-
-        # Calculate the rank to recv by this rank
-        num_dst_per_sender = len(ranks_to_recv) // len(ranks_to_send)
-        recver_pos = ranks_to_recv.index(ep_rank)
-        remainder_start = len(ranks_to_send) * num_dst_per_sender
-        if recver_pos < remainder_start:
-            src = ranks_to_send[recver_pos // num_dst_per_sender]
-        else:
-            src = ranks_to_send[recver_pos - remainder_start]
-
-        src_global = get_global_rank(ep_group, src)
-        p2p_ops += [
-            P2POp(
-                torch.distributed.irecv,
-                weight[dst],
-                src_global,
-            )
-            for weight in expert_weights_buffer
-        ]
-
     # 4. Execute the P2P operations. The real communication happens here.
     if p2p_ops and cuda_stream is not None:
         with torch.cuda.stream(cuda_stream):
@@ -237,38 +342,95 @@ def move_to_buffer(
         for req in reqs:
             req.wait()
     # wait for the communication to finish
-    return is_unchanged, is_received_locally, experts_recv_loc
+    return (
+        is_unchanged,
+        is_received_locally,
+        RecvMetadata(
+            recv_primary_mask=recv_primary_mask,
+            recv_count=recv_count,
+            recv_expert_ids=recv_expert_ids,
+            recv_dst_rows=recv_dst_rows,
+        ),
+    )
 
 
 def move_from_buffer(
     expert_weights: Iterable[torch.Tensor],
-    expert_weights_buffer: list[torch.Tensor],
+    expert_weights_buffers: list[torch.Tensor],
-    is_unchanged: list[bool],
+    is_unchanged: np.ndarray,
-    is_received_locally: list[bool],
+    is_received_locally: np.ndarray,
-    experts_recv_loc: dict[int, int],
+    recv_metadata: RecvMetadata,
-    new_indices: Sequence[int],
+    new_indices: np.ndarray,
-    ep_group: ProcessGroup,
+    ep_rank: int,
 ) -> None:
-    ep_rank = ep_group.rank()
+    """
-    num_local_experts = len(is_unchanged)
+    Copies expert weights from communication buffers back to the target weight tensors
+    after EPLB rebalancing.
 
-    local2global = partial(
+    Args:
-        idx_local_to_global, local_cnt=num_local_experts, ep_rank=ep_rank
+        expert_weights: List of the actual MoE layer weights used in the execution.
+        expert_weights_buffers: Intermediate buffers containing the experts weights
+            after the transfer is completed.
+        is_unchanged: (num_local_experts,), True where an expert row is unchanged.
+        is_received_locally: (num_local_experts,), True where a row is updated locally.
+        recv_metadata: RecvMetadata containing remote receive metadata.
+        new_indices: (num_experts_total,) mapping from local rows to desired
+            (possibly global) expert id, after rebalance.
+        ep_rank: Rank of the process in the expert parallel group.
+    """
+    recv_primary_mask = recv_metadata.recv_primary_mask
+    recv_count = recv_metadata.recv_count
+    recv_expert_ids = recv_metadata.recv_expert_ids
+    recv_dst_rows = recv_metadata.recv_dst_rows
+    num_local_experts = is_unchanged.shape[0]
+
+    # Mask for rows to copy back from buffers:
+    # copy if locally received OR remote primary recv
+    copy_mask = np.logical_or(is_received_locally, recv_primary_mask)
+    dest_mask_np = np.logical_and(~is_unchanged, copy_mask)
+    if bool(dest_mask_np.any()):
+        dest_indices = np.nonzero(dest_mask_np)[0].tolist()
+        for dst in dest_indices:
+            for w, b in zip(expert_weights, expert_weights_buffers):
+                w[dst].copy_(b[dst], non_blocking=True)
+
+    if recv_count == 0:
+        return
+
+    # Duplicate remote received rows to non-primary duplicate dsts
+    base = ep_rank * num_local_experts
+    local_experts = new_indices[base + np.arange(num_local_experts, dtype=np.int32)]
+    duplicate_mask = np.logical_and(
+        np.logical_and(~is_unchanged, ~is_received_locally),
+        np.logical_and(~recv_primary_mask, local_experts != -1),
     )
+    # All received experts are unique in the destination, so no need to copy duplicates
+    if not bool(duplicate_mask.any()):
+        return
 
-    for dst in range(num_local_experts):
+    dup_dst_rows = np.nonzero(duplicate_mask)[0]
-        if is_unchanged[dst]:
+    dup_experts = local_experts[dup_dst_rows]
-            continue
+
-        if is_received_locally[dst]:
+    prim_experts = recv_expert_ids[:recv_count]
-            for weight, buffer in zip(expert_weights, expert_weights_buffer):
+    prim_dsts = recv_dst_rows[:recv_count]
-                weight[dst].copy_(buffer[dst], non_blocking=True)
+    order = np.argsort(prim_experts, kind="stable")
-        else:
+    prim_experts_sorted = prim_experts[order]
-            expert = new_indices[local2global(dst)]
+    prim_dsts_sorted = prim_dsts[order]
-            if expert == -1:
+    pos = np.searchsorted(prim_experts_sorted, dup_experts)
-                continue
+    valid = np.logical_and(
-            src = experts_recv_loc[expert]
+        pos < prim_experts_sorted.shape[0],
-            for weight, buffer in zip(expert_weights, expert_weights_buffer):
+        prim_experts_sorted[np.minimum(pos, prim_experts_sorted.shape[0] - 1)]
-                weight[dst].copy_(buffer[src], non_blocking=True)
+        == dup_experts,
+    )
+    if not bool(valid.any()):
+        return
+
+    matched_dst_rows = dup_dst_rows[valid]
+    matched_src_rows = prim_dsts_sorted[pos[valid]]
+
+    for dst, src in zip(matched_dst_rows.tolist(), matched_src_rows.tolist()):
+        for w in expert_weights:
+            w[dst].copy_(w[src], non_blocking=True)
 
 
 async def transfer_layer(
@@ -281,7 +443,7 @@ async def transfer_layer(
     layer: int = 0,
     cuda_stream: torch.cuda.Stream | None = None,
     rank_mapping: dict[int, int] | None = None,
-) -> tuple[list[bool], list[bool], dict[int, int]]:
+) -> MoveToBufferResult:
     """
     Rearranges the expert weights in place according to the new expert indices.
 
@@ -299,6 +461,13 @@ async def transfer_layer(
         is_profile (bool): If `True`, do not perform any actual weight copy.
             This is used during profile run, where we only perform dummy
             communications to reserve enough memory for the buffers.
+
+    Returns:
+        is_unchanged (np.ndarray): (1, num_local_experts), True where expert
+            is left unchanged.
+        is_received_locally (np.ndarray): (1, num_local_experts), True where expert
+            can be received locally.
+        RecvMetadata: Metadata needed for completing remote weight transfers.
     """
     ep_size = ep_group.size()
     if rank_mapping is not None:
@@ -323,16 +492,19 @@ async def transfer_layer(
     assert new_global_expert_indices.shape == (num_moe_layers, num_physical_experts)
     assert num_physical_experts == ep_size * num_local_physical_experts
 
-    is_unchanged, is_received_locally, experts_recv_loc = move_to_buffer(
+    old_global_expert_indices_np = old_global_expert_indices.cpu().numpy()
+    new_global_expert_indices_np = new_global_expert_indices.cpu().numpy()
+
+    is_unchanged, is_received_locally, recv_metadata = move_to_buffer(
         num_local_experts=num_local_physical_experts,
-        old_indices=old_global_expert_indices[layer].tolist(),
+        old_indices=old_global_expert_indices_np[layer],
-        new_indices=new_global_expert_indices[layer].tolist(),
+        new_indices=new_global_expert_indices_np[layer],
         expert_weights=expert_weights[layer],
-        expert_weights_buffer=expert_weights_buffer,
+        expert_weights_buffers=expert_weights_buffer,
         cuda_stream=cuda_stream,
         ep_group=ep_group,
     )
-    return is_unchanged, is_received_locally, experts_recv_loc
+    return is_unchanged, is_received_locally, recv_metadata
 
 
 def rearrange_expert_weights_inplace(
@@ -388,19 +560,17 @@ def rearrange_expert_weights_inplace(
     ep_size = ep_group.size()
     assert num_physical_experts == ep_size * num_local_physical_experts
 
-    # A buffer to hold the expert weights in one layer during the exchange.
+    first_layer_weights = list(expert_weights[0])
+    # Buffers to hold the expert weights during the exchange.
     # NOTE: Currently we assume the same weights across different layers
     # have the same shape.
-    expert_weights_buffer = [torch.empty_like(w) for w in expert_weights[0]]
+    weights_buffer: list[torch.Tensor] = [
-
+        torch.empty_like(w) for w in first_layer_weights
+    ]
     if is_profile:
-        # Maximum send size is to send all local experts to all ranks,
+        # Reserve communication buffers via a minimal dummy all_gather on first layer
-        # So we use a dummy `all_gather` to reserve enough communication buffer
+        for weight, buffer in zip(expert_weights[0], weights_buffer):
-        for weight, buffer in zip(expert_weights[0], expert_weights_buffer):
-            # A `/dev/null`-like buffer to avoid real memory allocation
             dummy_recv_buffer = [buffer for _ in range(ep_size)]
-            # NOTE(bowen): Needed this barrier to avoid OOM during actual
-            # execution. I'm not very sure why this is needed
             torch.distributed.barrier()
             all_gather(
                 dummy_recv_buffer,
@@ -409,32 +579,32 @@ def rearrange_expert_weights_inplace(
             )
         return
 
-    old_global_expert_indices_cpu = old_global_expert_indices.cpu()
-    new_global_expert_indices_cpu = new_global_expert_indices.cpu()
-
     # NOTE(bowen): We need this synchronize to run, but I don't know why.
     # If you figure out the reason, please let me know -- thank you!
     torch.cuda.synchronize()
 
-    for layer in range(num_moe_layers):
+    old_global_expert_indices_cpu = old_global_expert_indices.cpu().numpy()
-        is_unchanged, is_received_locally, experts_recv_loc = move_to_buffer(
+    new_global_expert_indices_cpu = new_global_expert_indices.cpu().numpy()
+
+    for layer_idx in range(num_moe_layers):
+        is_unchanged, is_received_locally, recv_metadata = move_to_buffer(
             num_local_experts=num_local_physical_experts,
-            old_indices=old_global_expert_indices_cpu[layer].tolist(),
+            old_indices=old_global_expert_indices_cpu[layer_idx],
-            new_indices=new_global_expert_indices_cpu[layer].tolist(),
+            new_indices=new_global_expert_indices_cpu[layer_idx],
-            expert_weights=expert_weights[layer],
+            expert_weights=expert_weights[layer_idx],
-            expert_weights_buffer=expert_weights_buffer,
+            expert_weights_buffers=weights_buffer,
             cuda_stream=None,
             ep_group=ep_group,
         )
 
         move_from_buffer(
-            expert_weights=expert_weights[layer],
+            expert_weights=expert_weights[layer_idx],
-            expert_weights_buffer=expert_weights_buffer,
+            expert_weights_buffers=weights_buffer,
             is_unchanged=is_unchanged,
             is_received_locally=is_received_locally,
-            experts_recv_loc=experts_recv_loc,
+            recv_metadata=recv_metadata,
-            new_indices=new_global_expert_indices[layer].tolist(),
+            new_indices=new_global_expert_indices_cpu[layer_idx],
-            ep_group=ep_group,
+            ep_rank=ep_group.rank(),
         )
 
 
@@ -526,4 +696,4 @@ def _map_new_expert_indices_with_rank_mapping(
     return mapped_expert_indices
 
 
-__all__ = ["transfer_layer", "move_from_buffer"]
+__all__ = ["transfer_layer", "move_from_buffer", "RecvMetadata"]