[Fix] fix flashinfer usage for window attention (#1107)

python/sglang/srt/layers/radix_attention.py

@@ -120,12 +120,9 @@ class RadixAttention(nn.Module):
         # using two wrappers is unnecessary in the current PR, but are prepared for future PRs
         prefill_wrapper_ragged = input_metadata.flashinfer_prefill_wrapper_ragged
         prefill_wrapper_paged = input_metadata.flashinfer_prefill_wrapper_paged
-        if self.sliding_window_size != -1:
-            prefill_wrapper_ragged = prefill_wrapper_ragged[0]
+        if self.sliding_window_size != -1 or self.reuse:
             prefill_wrapper_paged = prefill_wrapper_paged[0]
         else:
-            if isinstance(prefill_wrapper_ragged, list):
-                prefill_wrapper_ragged = prefill_wrapper_ragged[1]
             if isinstance(prefill_wrapper_paged, list):
                 prefill_wrapper_paged = prefill_wrapper_paged[1]
 

python/sglang/srt/model_executor/forward_batch_info.py

@@ -324,9 +324,11 @@ def update_flashinfer_indices(
     else:
         kv_last_page_len = torch.ones((batch_size,), dtype=torch.int32, device="cuda")
         for wrapper_id in range(2):
-            if flashinfer_use_ragged:
+            if flashinfer_use_ragged and wrapper_id == 1:
+                # full attention use ragged+paged
                 paged_kernel_lens = prefix_lens
             else:
+                # window attention use paged only
                 paged_kernel_lens = seq_lens
 
             if wrapper_id == 0 and forward_mode == ForwardMode.DECODE:
@@ -374,13 +376,9 @@ def update_flashinfer_indices(
                 )
                 qo_indptr[1:] = torch.cumsum(seq_lens - prefix_lens, dim=0)
 
-                if flashinfer_use_ragged:
-                    model_runner.flashinfer_prefill_wrapper_ragged[
-                        wrapper_id
-                    ].end_forward()
-                    model_runner.flashinfer_prefill_wrapper_ragged[
-                        wrapper_id
-                    ].begin_forward(
+                if flashinfer_use_ragged and wrapper_id == 1:
+                    model_runner.flashinfer_prefill_wrapper_ragged.end_forward()
+                    model_runner.flashinfer_prefill_wrapper_ragged.begin_forward(
                         qo_indptr,
                         qo_indptr,
                         num_qo_heads,

python/sglang/srt/model_executor/model_runner.py

@@ -342,15 +342,14 @@ class ModelRunner:
                 dtype=torch.uint8,
                 device="cuda",
             )
-            self.flashinfer_prefill_wrapper_ragged = []
+            self.flashinfer_prefill_wrapper_ragged = (
+                BatchPrefillWithRaggedKVCacheWrapper(
+                    self.flashinfer_workspace_buffer, "NHD"
+                )
+            )
             self.flashinfer_prefill_wrapper_paged = []
             self.flashinfer_decode_wrapper = []
             for i in range(2):
-                self.flashinfer_prefill_wrapper_ragged.append(
-                    BatchPrefillWithRaggedKVCacheWrapper(
-                        self.flashinfer_workspace_buffer, "NHD"
-                    )
-                )
                 self.flashinfer_prefill_wrapper_paged.append(
                     BatchPrefillWithPagedKVCacheWrapper(
                         self.flashinfer_workspace_buffer, "NHD"