[Perf] Optimize compute maxsim using batched version, 3.2% E2E throughput improvement (#36710)

Signed-off-by: yewentao256 <zhyanwentao@126.com>

vllm/v1/pool/late_interaction.py

@@ -2,6 +2,7 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
 import zlib
+from collections.abc import Sequence
 
 import torch
 
@@ -62,3 +63,81 @@ def compute_maxsim_score(
     # compute in float32 for numerical stability
     token_scores = torch.matmul(q_emb.float(), d_emb.float().T)
     return token_scores.amax(dim=-1).sum()
+
+
+def compute_maxsim_scores(
+    q_embs: Sequence[torch.Tensor],
+    d_embs: Sequence[torch.Tensor],
+    max_batch_size: int = 64,
+    max_score_matrix_elements: int = 64_000_000,
+) -> list[torch.Tensor]:
+    """Compute MaxSim for multiple query/doc pairs in mini-batches."""
+    if len(q_embs) != len(d_embs):
+        raise ValueError("q_embs and d_embs must have the same length")
+
+    num_pairs = len(q_embs)
+    if num_pairs == 0:
+        return []
+
+    if max_batch_size <= 0:
+        raise ValueError("max_batch_size must be greater than 0")
+    if max_score_matrix_elements <= 0:
+        raise ValueError("max_score_matrix_elements must be greater than 0")
+
+    for q_emb, d_emb in zip(q_embs, d_embs):
+        if q_emb.ndim != 2 or d_emb.ndim != 2:
+            raise ValueError("Each embedding tensor must be 2-D")
+        if q_emb.shape[1] != d_emb.shape[1]:
+            raise ValueError("Query and document embeddings must have same dim")
+        if q_emb.device != d_emb.device:
+            raise ValueError("Query and document embeddings must be on same device")
+
+    scores: list[torch.Tensor] = []
+    start = 0
+    while start < num_pairs:
+        end = min(start + max_batch_size, num_pairs)
+        max_q = max(int(x.shape[0]) for x in q_embs[start:end])
+        max_d = max(int(x.shape[0]) for x in d_embs[start:end])
+
+        # keep score matrix bounded to avoid oversized allocations.
+        while (
+            end - start > 1
+            and (end - start) * max_q * max_d > max_score_matrix_elements
+        ):
+            end -= 1
+            max_q = max(int(x.shape[0]) for x in q_embs[start:end])
+            max_d = max(int(x.shape[0]) for x in d_embs[start:end])
+
+        batch_q = q_embs[start:end]
+        batch_d = d_embs[start:end]
+        batch_size = end - start
+        device = batch_q[0].device
+        dim = int(batch_q[0].shape[1])
+
+        q_batch = torch.zeros(
+            (batch_size, max_q, dim), dtype=torch.float32, device=device
+        )
+        d_batch = torch.zeros(
+            (batch_size, max_d, dim), dtype=torch.float32, device=device
+        )
+        q_mask = torch.zeros((batch_size, max_q), dtype=torch.bool, device=device)
+        d_mask = torch.zeros((batch_size, max_d), dtype=torch.bool, device=device)
+
+        # copy to padded tensors
+        for i, (q_emb, d_emb) in enumerate(zip(batch_q, batch_d)):
+            q_len = int(q_emb.shape[0])
+            d_len = int(d_emb.shape[0])
+            q_batch[i, :q_len] = q_emb.to(device=device, dtype=torch.float32)
+            d_batch[i, :d_len] = d_emb.to(device=device, dtype=torch.float32)
+            q_mask[i, :q_len] = True
+            d_mask[i, :d_len] = True
+
+        token_scores = torch.bmm(q_batch, d_batch.transpose(1, 2))
+        token_scores.masked_fill_(~d_mask.unsqueeze(1), float("-inf"))
+        max_per_query = token_scores.amax(dim=-1)
+        max_per_query.masked_fill_(~q_mask, 0.0)
+        batch_scores = max_per_query.sum(dim=-1)
+        scores.extend(batch_scores.unbind(0))
+        start = end
+
+    return scores

vllm/v1/worker/gpu/pool/late_interaction_runner.py

@@ -9,7 +9,7 @@ from vllm.v1.outputs import PoolerOutput
 from vllm.v1.pool.late_interaction import (
     LATE_INTERACTION_MODE_CACHE_QUERY,
     LATE_INTERACTION_MODE_SCORE_DOC,
-    compute_maxsim_score,
+    compute_maxsim_scores,
 )
 
 
@@ -72,6 +72,11 @@ class LateInteractionRunner:
             return raw_pooler_output
 
         outputs: list[torch.Tensor | None] = list(raw_pooler_output)
+        score_indices: list[int] = []
+        score_req_ids: list[str] = []
+        score_query_keys: list[str] = []
+        score_queries: list[torch.Tensor] = []
+        score_docs: list[torch.Tensor] = []
         for i, (req_id, output, params, finished) in enumerate(
             zip(req_ids, outputs, pooling_params, finished_mask)
         ):
@@ -101,13 +106,24 @@ class LateInteractionRunner:
                         "before their paired document requests."
                     )
 
-                outputs[i] = compute_maxsim_score(query_output, output)
-                self._doc_query_keys.pop(req_id, None)
-                self._release_query_use(query_key)
+                score_indices.append(i)
+                score_req_ids.append(req_id)
+                score_query_keys.append(query_key)
+                score_queries.append(query_output)
+                score_docs.append(output)
                 continue
 
             raise ValueError(f"Unsupported late-interaction mode: {mode!r}")
 
+        if score_indices:
+            score_values = compute_maxsim_scores(score_queries, score_docs)
+            for i, req_id, query_key, score in zip(
+                score_indices, score_req_ids, score_query_keys, score_values
+            ):
+                outputs[i] = score
+                self._doc_query_keys.pop(req_id, None)
+                self._release_query_use(query_key)
+
         return outputs
 
     def _release_query_use(self, query_key: str) -> None: