[ROCm] [Bugfix] Fix DeepSeek V4 Functionality and Accuracy (#42810)
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
This commit is contained in:
@@ -505,27 +505,6 @@ class SparseAttnIndexer(CustomOp):
|
||||
assert isinstance(q_quant, torch.Tensor), (
|
||||
"AMD sparse_attn_indexer expects a single FP8 q_quant tensor"
|
||||
)
|
||||
if self.skip_k_cache_insert or not rocm_aiter_ops.is_enabled():
|
||||
from vllm.v1.attention.ops.rocm_aiter_mla_sparse import (
|
||||
rocm_aiter_sparse_attn_indexer_native,
|
||||
)
|
||||
|
||||
return rocm_aiter_sparse_attn_indexer_native(
|
||||
hidden_states,
|
||||
_encode_layer_name(self.k_cache.prefix),
|
||||
self.k_cache.kv_cache,
|
||||
q_quant,
|
||||
k,
|
||||
weights,
|
||||
self.quant_block_size,
|
||||
self.scale_fmt,
|
||||
self.topk_tokens,
|
||||
self.head_dim,
|
||||
self.max_model_len,
|
||||
self.max_total_seq_len,
|
||||
self.topk_indices_buffer,
|
||||
skip_k_cache_insert=self.skip_k_cache_insert,
|
||||
)
|
||||
if rocm_aiter_ops.is_enabled():
|
||||
return torch.ops.vllm.rocm_aiter_sparse_attn_indexer(
|
||||
hidden_states,
|
||||
@@ -541,5 +520,9 @@ class SparseAttnIndexer(CustomOp):
|
||||
self.max_model_len,
|
||||
self.max_total_seq_len,
|
||||
self.topk_indices_buffer,
|
||||
skip_k_cache_insert=self.skip_k_cache_insert,
|
||||
)
|
||||
raise RuntimeError("Sparse attention indexer ROCm path could not be selected.")
|
||||
raise RuntimeError(
|
||||
"Sparse attention indexer ROCm path is only supported on AITER. "
|
||||
"Please enable aiter with VLLM_ROCM_USE_AITER=1"
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user