implement mla for fa3

Author: Fridge003

python/sglang/srt/layers/attention/flashattention_backend.py

@@ -13,7 +13,9 @@
 
 import torch
 
+from sglang.srt.configs.model_config import AttentionArch
 from sglang.srt.layers.attention.base_attn_backend import AttentionBackend
+from sglang.srt.managers.schedule_batch import global_server_args_dict
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch, ForwardMode
 
 if TYPE_CHECKING:
@@ -57,7 +59,9 @@ def __init__(
         self.device = model_runner.device
         self.decode_cuda_graph_metadata = {}
         self.req_to_token = model_runner.req_to_token_pool.req_to_token
-        self.page_size = model_runner.page_size
+        self.use_mla = (
+            model_runner.model_config.attention_arch == AttentionArch.MLA
+        ) and (not global_server_args_dict["disable_mla"])
 
     def init_forward_metadata(self, forward_batch: ForwardBatch):
         """Initialize forward metadata to cache repetitive calculations."""
@@ -79,17 +83,6 @@ def init_forward_metadata(self, forward_batch: ForwardBatch):
         metadata.page_table = forward_batch.req_to_token_pool.req_to_token[
             forward_batch.req_pool_indices, : metadata.max_seq_len_k
         ]
-
-        # Precompute strided indices
-        # [0, page_size, 2 * page_size, ...]
-        if self.page_size > 1:
-            self.strided_indices = torch.arange(
-                0, metadata.page_table.shape[1], self.page_size, device=self.device
-            )
-            metadata.page_table = (
-                metadata.page_table[:, self.strided_indices] // self.page_size
-            )
-
         if forward_batch.forward_mode == ForwardMode.DECODE:
             # Precompute cumulative sequence lengths
             metadata.cu_seqlens_q = torch.arange(
@@ -117,23 +110,28 @@ def forward_extend(
         forward_batch: ForwardBatch,
         save_kv_cache=True,
     ):
-        cache_loc = (
-            forward_batch.out_cache_loc
-            if not layer.is_cross_attention
-            else forward_batch.encoder_out_cache_loc
-        )
 
-        if k is not None:
-            assert v is not None
-            if save_kv_cache:
+        if k is not None and v is not None and save_kv_cache:
+            cache_loc = (
+                forward_batch.out_cache_loc
+                if not layer.is_cross_attention
+                else forward_batch.encoder_out_cache_loc
+            )
+            if not self.use_mla:
                 forward_batch.token_to_kv_pool.set_kv_buffer(
                     layer, cache_loc, k, v, layer.k_scale, layer.v_scale
                 )
+            else:
+                forward_batch.token_to_kv_pool.set_kv_buffer(
+                    layer,
+                    cache_loc,
+                    k,
+                    v,
+                )
 
         # Use precomputed metadata
         metadata = self.forward_metadata
 
-        # # Use Flash Attention for prefill
         # Calculate window size (can be moved to metadata if layer properties don't change)
         # we don't do layer.sliding_window_size - 1 since in model.get_attention_sliding_window_size() we already - 1
         # here is two side inclusive
@@ -142,36 +140,55 @@ def forward_extend(
             if layer.sliding_window_size is not None
             else (-1, -1)
         )
-        kv_cache = forward_batch.token_to_kv_pool.get_kv_buffer(layer.layer_id)
-        key_cache, value_cache = kv_cache[0], kv_cache[1]
 
-        key_cache = key_cache.view(
-            -1, self.page_size, layer.tp_k_head_num, layer.head_dim
-        )
-        value_cache = value_cache.view(
-            -1, self.page_size, layer.tp_v_head_num, layer.head_dim
-        )
-
-        page_table = metadata.page_table
-
-        o = flash_attn_with_kvcache(
-            q=q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim),
-            k_cache=key_cache,
-            v_cache=value_cache,
-            page_table=page_table,
-            cache_seqlens=metadata.cache_seqlens_int32,
-            cu_seqlens_q=metadata.cu_seqlens_q,
-            cu_seqlens_k_new=metadata.cu_seqlens_k,
-            max_seqlen_q=metadata.max_seq_len_q,
-            softmax_scale=layer.scaling,
-            causal=True,
-            window_size=window_size,
-            softcap=layer.logit_cap,
-            k_descale=layer.k_scale,
-            v_descale=layer.v_scale,
-        )
+        # # Use Flash Attention for prefill
+        if not self.use_mla:
+            # Do multi-head attention
+            kv_cache = forward_batch.token_to_kv_pool.get_kv_buffer(layer.layer_id)
+            key_cache, value_cache = kv_cache[0], kv_cache[1]
+            o = flash_attn_with_kvcache(
+                q=q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim),
+                k_cache=key_cache.unsqueeze(1),
+                v_cache=value_cache.unsqueeze(1),
+                page_table=metadata.page_table,
+                cache_seqlens=metadata.cache_seqlens_int32,
+                cu_seqlens_q=metadata.cu_seqlens_q,
+                cu_seqlens_k_new=metadata.cu_seqlens_k,
+                max_seqlen_q=metadata.max_seq_len_q,
+                softmax_scale=layer.scaling,
+                causal=True,
+                window_size=window_size,
+                softcap=layer.logit_cap,
+                k_descale=layer.k_scale,
+                v_descale=layer.v_scale,
+            )
+        else:
+            # Do absorbed multi-latent attention
+            kv_cache = forward_batch.token_to_kv_pool.get_key_buffer(layer.layer_id)
+            c_kv = kv_cache[:, :, : layer.v_head_dim]
+            k_rope = kv_cache[:, :, layer.v_head_dim :]
+
+            q_all = q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim)
+            q_nope = q_all[:, :, : layer.v_head_dim]
+            q_rope = q_all[:, :, layer.v_head_dim :]
+            o = flash_attn_with_kvcache(
+                q=q_rope,
+                k_cache=k_rope.unsqueeze(1),
+                v_cache=c_kv.unsqueeze(1),
+                qv=q_nope,
+                page_table=metadata.page_table,
+                cache_seqlens=metadata.cache_seqlens_int32,
+                cu_seqlens_q=metadata.cu_seqlens_q,
+                cu_seqlens_k_new=metadata.cu_seqlens_k,
+                max_seqlen_q=metadata.max_seq_len_q,
+                softmax_scale=layer.scaling,
+                causal=True,
+                softcap=layer.logit_cap,
+                k_descale=layer.k_scale,
+                v_descale=layer.v_scale,
+            )
 
-        return o.view(-1, layer.tp_q_head_num * layer.head_dim)
+        return o.view(-1, layer.tp_q_head_num * layer.v_head_dim)
 
     def forward_decode(
         self,
@@ -190,18 +207,21 @@ def forward_decode(
                 if not layer.is_cross_attention
                 else forward_batch.encoder_out_cache_loc
             )
-            forward_batch.token_to_kv_pool.set_kv_buffer(
-                layer, cache_loc, k, v, layer.k_scale, layer.v_scale
-            )
+            if not self.use_mla:
+                forward_batch.token_to_kv_pool.set_kv_buffer(
+                    layer, cache_loc, k, v, layer.k_scale, layer.v_scale
+                )
+            else:
+                forward_batch.token_to_kv_pool.set_kv_buffer(
+                    layer,
+                    cache_loc,
+                    k,
+                    v,
+                )
 
-        # Get KV cache
-        kv_cache = forward_batch.token_to_kv_pool.get_kv_buffer(layer.layer_id)
-        key_cache, value_cache = kv_cache[0], kv_cache[1]
         # Use precomputed metadata
         metadata = self.forward_metadata
 
-        # Pre-reshape query tensor
-        q_reshaped = q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim)
         # Calculate window size (can be moved to metadata if layer properties don't change)
         # we don't do layer.sliding_window_size - 1 since in model.get_attention_sliding_window_size() we already - 1
         # here is two side inclusive
@@ -210,33 +230,62 @@ def forward_decode(
             if layer.sliding_window_size is not None
             else (-1, -1)
         )
-        # Run attention with precomputed values
-        key_cache = key_cache.view(
-            -1, self.page_size, layer.tp_k_head_num, layer.head_dim
-        )
-        value_cache = value_cache.view(
-            -1, self.page_size, layer.tp_v_head_num, layer.head_dim
-        )
 
-        page_table = metadata.page_table
-
-        o = flash_attn_with_kvcache(
-            q=q_reshaped,
-            k_cache=key_cache,
-            v_cache=value_cache,
-            page_table=page_table,
-            cache_seqlens=metadata.cache_seqlens_int32,
-            cu_seqlens_q=metadata.cu_seqlens_q,
-            cu_seqlens_k_new=metadata.cu_seqlens_k,
-            max_seqlen_q=1,
-            softmax_scale=layer.scaling,
-            causal=True,
-            window_size=window_size,
-            softcap=layer.logit_cap,
-            k_descale=layer.k_scale,
-            v_descale=layer.v_scale,
-        )
-        return o.view(-1, layer.tp_q_head_num * layer.head_dim)
+        if not self.use_mla:
+            # Do multi-head attention
+
+            # Get KV cache
+            kv_cache = forward_batch.token_to_kv_pool.get_kv_buffer(layer.layer_id)
+            key_cache, value_cache = kv_cache[0], kv_cache[1]
+
+            # Pre-reshape query tensor
+            q_reshaped = q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim)
+
+            # Run attention with precomputed values
+            o = flash_attn_with_kvcache(
+                q=q_reshaped,
+                k_cache=key_cache.unsqueeze(1),
+                v_cache=value_cache.unsqueeze(1),
+                page_table=metadata.page_table,
+                cache_seqlens=metadata.cache_seqlens_int32,
+                cu_seqlens_q=metadata.cu_seqlens_q,
+                cu_seqlens_k_new=metadata.cu_seqlens_k,
+                max_seqlen_q=1,
+                softmax_scale=layer.scaling,
+                causal=True,
+                window_size=window_size,
+                softcap=layer.logit_cap,
+                k_descale=layer.k_scale,
+                v_descale=layer.v_scale,
+            )
+        else:
+            # Do absorbed multi-latent attention
+            kv_cache = forward_batch.token_to_kv_pool.get_key_buffer(layer.layer_id)
+            c_kv = kv_cache[:, :, : layer.v_head_dim]
+            k_rope = kv_cache[:, :, layer.v_head_dim :]
+
+            q_all = q.contiguous().view(-1, layer.tp_q_head_num, layer.head_dim)
+            q_nope = q_all[:, :, : layer.v_head_dim]
+            q_rope = q_all[:, :, layer.v_head_dim :]
+
+            o = flash_attn_with_kvcache(
+                q=q_rope,
+                k_cache=k_rope.unsqueeze(1),
+                v_cache=c_kv.unsqueeze(1),
+                qv=q_nope,
+                page_table=metadata.page_table,
+                cache_seqlens=metadata.cache_seqlens_int32,
+                cu_seqlens_q=metadata.cu_seqlens_q,
+                cu_seqlens_k_new=metadata.cu_seqlens_k,
+                max_seqlen_q=1,
+                softmax_scale=layer.scaling,
+                causal=True,
+                softcap=layer.logit_cap,
+                k_descale=layer.k_scale,
+                v_descale=layer.v_scale,
+            )
+
+        return o.view(-1, layer.tp_q_head_num * layer.v_head_dim)
 
     def init_cuda_graph_state(self, max_bs: int):
         """Initialize CUDA graph state for the attention backend.
@@ -251,13 +300,7 @@ def init_cuda_graph_state(self, max_bs: int):
         self.decode_cuda_graph_metadata = {
             # Page table for token mapping (batch_size, max_context_len)
             "page_table": torch.zeros(
-                max_bs,
-                (self.max_context_len + self.page_size - 1) // self.page_size,
-                dtype=torch.int32,
-                device=self.device,
-            ),
-            "strided_indices": torch.arange(
-                0, self.max_context_len, self.page_size, device=self.device
+                max_bs, self.max_context_len, dtype=torch.int32, device=self.device
             ),
         }
 
@@ -286,7 +329,6 @@ def init_forward_metadata_capture_cuda_graph(
         metadata.page_table = self.decode_cuda_graph_metadata["page_table"][
             req_pool_indices, :
         ]
-
         if forward_mode == ForwardMode.DECODE:
             # Precompute cumulative sequence lengths
             metadata.cu_seqlens_q = torch.arange(