[main][bugfix] Fix bugs and refactor cached mask generation logic

Signed-off-by: rjg-lyh <[email protected]>

Author: rjg-lyh

tests/ut/attention/test_attention_mask.py

@@ -77,34 +77,22 @@ def test_get_splitfuse_attn_mask(self):
         attention_mask_builder = AttentionMaskBuilder(max_seq_len=1024,
                                                       dtype=torch.float16)
         attn_mask = attention_mask_builder.get_splitfuse_attn_mask(
-            seq_lens=[512],
-            query_lens=[512],
-            position=torch.tensor([0]),
+            seq_lens=torch.tensor([10, 20, 100]),
+            position=torch.tensor([7, 8, 9, 18, 19, 99]),
             dtype=torch.float16,
             device=torch.device("cpu"),
         )
-        self.assertEqual(attn_mask.shape, (1, 512))
+        self.assertEqual(attn_mask.shape, (6, 100))
         self.assertEqual(attention_mask_builder._seq_len_cached, 1024)
 
         attn_mask = attention_mask_builder.get_splitfuse_attn_mask(
-            seq_lens=[2048],
-            query_lens=[1024],
-            position=torch.tensor([0]),
+            seq_lens=torch.tensor([10, 3000, 2000]),
+            position=torch.tensor([7, 8, 9, 2999, 1999]),
             dtype=torch.float16,
             device=torch.device("cpu"),
         )
-        self.assertEqual(attn_mask.shape, (1024, 2048))
-
-        attention_mask_builder = AttentionMaskBuilder(max_seq_len=1024,
-                                                      dtype=torch.int8)
-        attn_mask = attention_mask_builder.get_splitfuse_attn_mask(
-            seq_lens=[512],
-            query_lens=[512],
-            position=torch.tensor([0]),
-            dtype=torch.int8,
-            device=torch.device("cpu"),
-        )
-        self.assertEqual(attn_mask.shape, (1, 512))
+        self.assertEqual(attn_mask.shape, (5, 3000))
+        self.assertEqual(attention_mask_builder._seq_len_cached, 3000)
 
     def test_use_multiple_masks(self):
         max_seq_lens = [128, 512, 1024]

vllm_ascend/attention/attention_mask.py

@@ -44,61 +44,41 @@ def __init__(
 
         self._seq_len_cached = attn_mask.shape[0]
         self.attn_mask_cache = attn_mask
-        self.splitfuse_mask_value = -10000
+
+    @staticmethod
+    def get_mask_scale_factor(dtype: torch.dtype = torch.float16):
+        mask_scale_factor = 1
+        if dtype == torch.bfloat16:
+            mask_scale_factor = -10000
+        return mask_scale_factor
 
     def get_attn_mask(self, max_seq_len: int, dtype: torch.dtype,
                       device: torch.device):
-        self._update_attn_cache(max_seq_len, dtype, device)
-        return self.attn_mask_cache[:max_seq_len, :max_seq_len].contiguous()
+        self._update_attn_cache(max_seq_len, dtype)
+        return self.attn_mask_cache[:max_seq_len, :max_seq_len].contiguous(
+        ).to(device)
 
     def get_splitfuse_attn_mask(
         self,
         seq_lens,
-        query_lens,
         position,
         dtype,
         device,
     ) -> torch.Tensor:
         max_seq_len = max(seq_lens, default=0)
-        if max_seq_len <= self._seq_len_cached:
-            self._update_attn_cache(max_seq_len, dtype, device)
-            # FIXME: Currently the mask value of chunked-prefill situation and Prefill-Only situation
-            # is not the same. Fix this in the future when kernel is ready.
-            if self.attn_mask_cache.numel(
-            ) > 1 and self.attn_mask_cache[0][1] > 0:
-                attn_mask = self.get_attn_mask(  # type: ignore
-                    max_seq_len, dtype, device)
-                # Do not use in-place multiplication to avoid modifying `self.attn_mask_cache`!
-                attn_mask = attn_mask * -10000
-            else:
-                attn_mask = self.attn_mask_cache
-            return torch.index_select(attn_mask, dim=0,
-                                      index=position)[:, :max_seq_len]
-        total_q_len = sum(query_lens)
-        attn_mask = torch.zeros((total_q_len, max_seq_len),
-                                dtype=dtype,
-                                device="cpu")
-        current_row = 0
-        for i in range(len(query_lens)):
-            seq_len = seq_lens[i]
-            q_len = query_lens[i]
-            context_len = seq_len - q_len
-
-            assert context_len >= 0
-            attn_mask[current_row:current_row + q_len,
-                      context_len:] = self.splitfuse_mask_value
-            right_tensor = attn_mask[current_row:current_row + q_len,
-                                     context_len:seq_len]
-            right_tensor.masked_fill_(
-                right_tensor.tril() == self.splitfuse_mask_value, 0)
-            current_row += q_len
-
-        return attn_mask.to(device, non_blocking=True)
+        self._update_attn_cache(max_seq_len, dtype)
+        # FIXME: Currently the mask value of chunked-prefill situation and Prefill-Only situation
+        # is not the same. Fix this in the future when kernel is ready.
+        mask_scale_factor = AttentionMaskBuilder.get_mask_scale_factor(dtype)
+        attn_mask = torch.index_select(self.attn_mask_cache,
+                                       dim=0,
+                                       index=position)[:, :max_seq_len]
+        attn_mask *= mask_scale_factor
+        return attn_mask.contiguous().to(device, non_blocking=True)
 
-    def _update_attn_cache(self, seqlen: int, dtype: torch.dtype,
-                           device: torch.device):
+    def _update_attn_cache(self, seqlen: int, dtype: torch.dtype):
         if seqlen > self._seq_len_cached:
             self._seq_len_cached = seqlen
             self.attn_mask_cache = _generate_attn_mask(seqlen, dtype)
-        if self.attn_mask_cache.device != device:
-            self.attn_mask_cache = self.attn_mask_cache.to(device)
+        if self.attn_mask_cache.dtype != dtype:
+            self.attn_mask_cache = self.attn_mask_cache.to(dtype)

vllm_ascend/worker/eagle_proposer_v1.py

@@ -79,11 +79,10 @@ def __init__(self,
     def _make_attention_mask(
         self,
         seq_lens,
-        query_lens,
         position,
     ) -> torch.Tensor:
         return self.attn_mask_builder.get_splitfuse_attn_mask(
-            seq_lens, query_lens, position, self.dtype, self.device)
+            seq_lens, position, self.dtype, self.device)
 
     def propose(
         self,
@@ -247,7 +246,6 @@ def propose(
             positions = positions_cpu.to(device)
             attn_mask = self._make_attention_mask(
                 seq_lens=attn_metadata.seq_lens,
-                query_lens=attn_metadata.max_query_len,
                 position=positions,
             )
             attn_metadata.attn_mask = attn_mask

vllm_ascend/worker/model_runner_v1.py

@@ -20,7 +20,6 @@
 import copy
 import gc
 import math
-import os
 import time
 import types
 from contextlib import contextmanager, nullcontext
@@ -228,8 +227,7 @@ def __init__(self, vllm_config: VllmConfig, device: torch.device):
         self.attn_metadata_builder = self.attn_backend.get_builder_cls()(
             vllm_config, device)
         self.attn_mask_builder = AttentionMaskBuilder(
-            min(self.model_config.max_model_len,
-                int(os.getenv("PAGED_ATTENTION_MASK_LEN", 10000))), self.dtype)
+            self.model_config.max_model_len, self.dtype)
 
         # Set up speculative decoding.
         self.use_aux_hidden_state_outputs = False
@@ -847,12 +845,12 @@ def get_supported_tasks(self) -> "tuple[SupportedTask, ...]":
 
         return tuple(tasks)
 
-    def _make_attention_mask(self, seq_lens, query_lens, position,
+    def _make_attention_mask(self, seq_lens, position,
                              attn_state) -> torch.Tensor:
         # Chunk Prefill situation.
         if attn_state == AscendAttentionState.ChunkedPrefill and not self.vllm_config.model_config.use_mla:
             return self.attn_mask_builder.get_splitfuse_attn_mask(
-                seq_lens, query_lens, position, self.dtype, self.device)
+                seq_lens, position, self.dtype, self.device)
         # Prefill without cache situation.
         elif attn_state == AscendAttentionState.PrefillNoCache:
             max_seq_len = max(seq_lens, default=0)
@@ -1124,16 +1122,17 @@ def _process_reqs(
                 self.mrope_positions_cpu[:, :total_num_scheduled_tokens],
                 non_blocking=True)
 
-        self.positions[total_num_scheduled_tokens:num_input_tokens].zero_()
-        self.positions[:total_num_scheduled_tokens].copy_(
-            self.positions_cpu[:total_num_scheduled_tokens], non_blocking=True)
+        self.positions_cpu[total_num_scheduled_tokens:num_input_tokens].zero_()
+        self.positions[:num_input_tokens].copy_(
+            self.positions_cpu[:num_input_tokens], non_blocking=True)
+        positions_cpu = self.positions_cpu[:num_input_tokens]
         positions = self.positions[:num_input_tokens]
         self.query_lens = torch.from_numpy(num_scheduled_tokens)
 
         self.seq_lens_np[:num_reqs] = (
             self.input_batch.num_computed_tokens_cpu[:num_reqs] +
             num_scheduled_tokens)
-        seq_lens = self.seq_lens_cpu[:num_reqs]
+        seq_lens_cpu = self.seq_lens_cpu[:num_reqs]
 
         block_table_indices = (req_indices * self.max_num_blocks_per_req +
                                positions_np // self.block_size)
@@ -1169,11 +1168,9 @@ def _process_reqs(
         else:
             attn_state = AscendAttentionState.PrefillCacheHit
 
-        self.attn_mask = self._make_attention_mask(
-            seq_lens=seq_lens,
-            query_lens=num_scheduled_tokens,
-            position=positions,
-            attn_state=attn_state)
+        self.attn_mask = self._make_attention_mask(seq_lens=seq_lens_cpu,
+                                                   position=positions_cpu,
+                                                   attn_state=attn_state)
         self.attn_state = attn_state  # type: ignore
 
         self.query_start_loc_np[0] = 0