add support for whisper v1 using aiter unified attention and aiter flash attention

Signed-off-by: apinge <Tong.Qiu2@amd.com>

Author: apinge

vllm/v1/attention/backends/rocm_aiter_fa.py

@@ -513,12 +513,9 @@ def __init__(
         assert self.num_heads % self.num_kv_heads == 0
         self.num_queries_per_kv = self.num_heads // self.num_kv_heads
 
-        if attn_type != AttentionType.DECODER:
+        if attn_type not in [AttentionType.DECODER, AttentionType.ENCODER_DECODER]:
             raise NotImplementedError(
-                "Encoder self-attention and "
-                "encoder/decoder cross-attention "
-                "are not implemented for "
-                "FlashAttentionImpl"
+                "Encoder self-attention is not implemented for FlashAttentionImpl"
             )
 
     def extend_forward(
@@ -674,7 +671,14 @@ def forward(
         # performance to make sure it does not introduce any overhead.
         num_actual_tokens = attn_metadata.num_actual_tokens
         key_cache, value_cache = kv_cache.unbind(0)
-        if self.kv_sharing_target_layer_name is None:
+        # key and value may be None in the case of cross attention. They are
+        # calculated once based on the output from the encoder and then cached
+        # in KV cache.
+        if (
+            self.kv_sharing_target_layer_name is None
+            and key is not None
+            and value is not None
+        ):
             # Reshape the input keys and values and store them in the cache.
             # Skip this if sharing KV cache with an earlier attention layer.
             # NOTE(woosuk): Here, key and value are padded while slot_mapping
@@ -700,8 +704,8 @@ def forward(
 
         # decode:extend:prefill
         query = query[:num_actual_tokens]
-        key = key[:num_actual_tokens]
-        value = value[:num_actual_tokens]
+        key = key[:num_actual_tokens] if key is not None else key_cache[:num_actual_tokens]
+        value = value[:num_actual_tokens] if value is not None else value_cache[:num_actual_tokens]
 
         output_actual_tokens = output[:num_actual_tokens]
 

vllm/v1/attention/backends/rocm_aiter_unified_attn.py

@@ -142,7 +142,14 @@ def forward(
 
         key_cache, value_cache = kv_cache.unbind(0)
 
-        if self.kv_sharing_target_layer_name is None:
+        # key and value may be None in the case of cross attention. They are
+        # calculated once based on the output from the encoder and then cached
+        # in KV cache.
+        if (
+            self.kv_sharing_target_layer_name is None
+            and key is not None
+            and value is not None
+        ):
             # Reshape the input keys and values and store them in the cache.
             # Skip this if sharing KV cache with an earlier attention layer.
             ops.reshape_and_cache_flash(
@@ -169,7 +176,7 @@ def forward(
         max_seqlen_k = attn_metadata.max_seq_len
         block_table = attn_metadata.block_table
 
-        descale_shape = (cu_seqlens_q.shape[0] - 1, key.shape[1])
+        descale_shape = (cu_seqlens_q.shape[0] - 1, key.shape[1] if key is not None else self.num_kv_heads)
 
         self.unified_attention(
             q=query[:num_actual_tokens],

vllm/v1/attention/backends/rocm_attn.py

@@ -238,12 +238,9 @@ def __init__(
 
         RocmAttentionBackend.validate_head_size(head_size)
 
-        if attn_type != AttentionType.DECODER:
+        if attn_type not in [AttentionType.DECODER, AttentionType.ENCODER_DECODER]:
             raise NotImplementedError(
-                "Encoder self-attention and "
-                "encoder/decoder cross-attention "
-                "are not implemented for "
-                "RocmAttentionImpl"
+                "Encoder self-attention is not implemented for RocmAttentionImpl"
             )
 
         self.fp8_dtype = current_platform.fp8_dtype()