EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Drop 0.11.0 support (#4377)

Browse Source 
There is a lot hack code for v0.11.0, which makes the code hard to
upgrade to newer vLLM version. Since v0.11.0 will release soon. Let's
drop v0.11.0 support first. Then we'll upgrade to v0.11.2 soon.


- vLLM version: v0.11.0
- vLLM main:
2918c1b49c

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-11-24 17:08:20 +08:00
committed by GitHub
parent 41ddb06554
commit a1f142b7ad
80 changed files with 467 additions and 1755 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
vllm_ascend/torchair/models/qwen3_moe.py
View File

@@ -23,7 +23,7 @@ from torch import nn
from transformers import PretrainedConfig
from vllm.attention import Attention, AttentionMetadata
from vllm.compilation.decorators import support_torch_compile
from vllm.config import CacheConfig, VllmConfig
from vllm.config import CacheConfig, CompilationMode, VllmConfig
from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size
from vllm.distributed.parallel_state import (get_dp_group, get_ep_group,
get_tp_group)
@@ -55,12 +55,6 @@ from vllm_ascend.attention.attention_v1 import AscendAttentionState
from vllm_ascend.torchair.ops.sequence_parallel import (MetadataForPadding,
init_metadata_for_sp)
from vllm_ascend.torchair.ops.torchair_fused_moe import TorchairAscendFusedMoE
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.config import CompilationLevel
else:
from vllm.config import CompilationMode
class CustomSparseMoeBlock(Qwen3MoeSparseMoeBlock):
@@ -299,16 +293,10 @@ class CustomQwen3MoeDecoderLayer(Qwen3MoeDecoderLayer):
layer_idx = extract_layer_index(prefix)
mlp_only_layers = ([] if not hasattr(config, "mlp_only_layers") else
config.mlp_only_layers)
if vllm_version_is("0.11.0"):
self.use_aclgraph = (vllm_config is not None
and vllm_config.compilation_config.level
== CompilationLevel.PIECEWISE and
not vllm_config.model_config.enforce_eager)
else:
self.use_aclgraph = (vllm_config is not None
and vllm_config.compilation_config.mode
== CompilationMode.VLLM_COMPILE and
not vllm_config.model_config.enforce_eager)
self.use_aclgraph = (vllm_config is not None
and vllm_config.compilation_config.mode
== CompilationMode.VLLM_COMPILE
and not vllm_config.model_config.enforce_eager)
if (layer_idx not in mlp_only_layers) and (
config.num_experts > 0 and
(layer_idx + 1) % config.decoder_sparse_step == 0):

178
vllm_ascend/torchair/models/torchair_deepseek_v2.py
View File

@@ -32,6 +32,7 @@ import torch_npu
from torch import nn
from transformers import PretrainedConfig
from vllm.attention import AttentionMetadata
from vllm.attention.layer import MLAAttention
from vllm.config import CacheConfig, ModelConfig, VllmConfig
from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank,
get_tensor_model_parallel_world_size,
@@ -74,12 +75,7 @@ from vllm_ascend.quantization.quant_config import AscendLinearMethod
from vllm_ascend.torchair.ops.torchair_fused_moe import TorchairAscendFusedMoE
from vllm_ascend.torchair.quantization.torchair_w8a8_dynamic import \
TorchairAscendW8A8DynamicLinearMethod
from vllm_ascend.utils import dispose_tensor, oproj_tp_enable, vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.attention import Attention
else:
from vllm.attention.layer import MLAAttention
from vllm_ascend.utils import dispose_tensor, oproj_tp_enable
class Indexer(nn.Module):
@@ -616,67 +612,31 @@ class TorchairDeepseekV2MLAAttention(DeepseekV2MLAAttention):
# k_c.size(1) + k_pe.size(1) == kv_cache.size(2)
# i.e.
# kv_lora_rank + qk_rope_head_dim == head_size
if vllm_version_is("0.11.0"):
self.mla_attn = Attention(
num_heads=self.num_local_heads,
head_size=self.kv_lora_rank + self.qk_rope_head_dim,
scale=self.scaling,
num_kv_heads=1,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_mla=True,
use_sparse=False,
indexer=None,
# SFA Args
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
qk_head_dim=self.qk_head_dim,
v_head_dim=self.v_head_dim,
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj
if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj
if self.q_lora_rank is None else self.q_b_proj,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
decoder_layer=decoder_layer,
)
else:
self.mla_attn = MLAAttention(
num_heads=self.num_local_heads,
scale=self.scaling,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
v_head_dim=self.v_head_dim,
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_sparse=False,
indexer=None,
# MLA Args
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj
if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj
if self.q_lora_rank is None else self.q_b_proj,
q_b_proj=self.q_b_proj
if self.q_lora_rank is not None else None,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
)
self.mla_attn = MLAAttention(
num_heads=self.num_local_heads,
scale=self.scaling,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
v_head_dim=self.v_head_dim,
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_sparse=False,
indexer=None,
# MLA Args
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj if self.q_lora_rank is None else self.q_b_proj,
q_b_proj=self.q_b_proj if self.q_lora_rank is not None else None,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
)
def forward(
self,
@@ -882,66 +842,30 @@ class TorchairDeepseekV2SFAAttention(DeepseekV2MLAAttention):
index_topk=self.index_topk,
prefix=f"{prefix}.indexer",
)
if vllm_version_is("0.11.0"):
self.sfa_attn = Attention(
num_heads=self.num_local_heads,
head_size=self.kv_lora_rank + self.qk_rope_head_dim,
scale=self.scaling,
num_kv_heads=1,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_mla=True,
use_sparse=True,
indexer=self.indexer,
# SFA Args
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
qk_head_dim=self.qk_head_dim,
v_head_dim=self.v_head_dim,
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj
if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj
if self.q_lora_rank is None else self.q_b_proj,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
decoder_layer=decoder_layer,
)
else:
self.sfa_attn = MLAAttention(
num_heads=self.num_local_heads,
scale=self.scaling,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
v_head_dim=self.v_head_dim,
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_sparse=True,
indexer=self.indexer,
# MLA Args
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj
if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj
if self.q_lora_rank is None else self.q_b_proj,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
)
self.sfa_attn = MLAAttention(
num_heads=self.num_local_heads,
scale=self.scaling,
qk_nope_head_dim=self.qk_nope_head_dim,
qk_rope_head_dim=self.qk_rope_head_dim,
v_head_dim=self.v_head_dim,
q_lora_rank=self.q_lora_rank,
kv_lora_rank=self.kv_lora_rank,
cache_config=cache_config,
quant_config=quant_config,
prefix=f"{prefix}.attn",
use_sparse=True,
indexer=self.indexer,
# MLA Args
rotary_emb=self.rotary_emb,
q_a_proj=self.q_a_proj if self.q_lora_rank is not None else None,
q_a_layernorm=self.q_a_layernorm
if self.q_lora_rank is not None else None,
q_proj=self.q_proj if self.q_lora_rank is None else self.q_b_proj,
kv_a_proj_with_mqa=self.kv_a_proj_with_mqa,
kv_a_layernorm=self.kv_a_layernorm,
kv_b_proj=self.kv_b_proj,
o_proj=self.o_proj,
)
def forward(
self,

11
vllm_ascend/torchair/ops/torchair_fused_moe.py
View File

@@ -53,8 +53,7 @@ from vllm_ascend.torchair.utils import (get_all_reduce_merge_state,
super_kernel)
from vllm_ascend.utils import (AscendSocVersion, dispose_tensor,
get_ascend_soc_version, is_310p,
is_hierarchical_communication_enabled,
vllm_version_is)
is_hierarchical_communication_enabled)
def torchair_fused_experts_with_mc2(
@@ -1069,12 +1068,8 @@ class TorchairAscendFusedMoE(FusedMoE):
get_compressed_expert_map(self.expert_map))
else:
# init moe.
if vllm_version_is("0.11.0"):
self.local_num_experts, self.expert_map = determine_expert_map(
self.ep_size, self.ep_rank, self.global_num_experts)
else:
self.local_num_experts, self.expert_map, _ = determine_expert_map(
self.ep_size, self.ep_rank, self.global_num_experts)
self.local_num_experts, self.expert_map, _ = determine_expert_map(
self.ep_size, self.ep_rank, self.global_num_experts)
# dynamic eplb initializing with not expert_map_path
if self.dynamic_eplb:
self.log2phy = determine_default_log2phy_map(

8
vllm_ascend/torchair/torchair_attention.py
View File

@@ -26,13 +26,7 @@ from vllm.attention.backends.abstract import (AttentionImpl, AttentionLayer,
AttentionType)
from vllm.attention.backends.utils import PAD_SLOT_ID
from vllm.config import VllmConfig
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import cdiv
else:
from vllm.utils.math_utils import cdiv
from vllm.utils.math_utils import cdiv
from vllm_ascend.attention.attention_v1 import (AscendAttentionBackend,
AscendAttentionMetadataBuilder,

8
vllm_ascend/torchair/torchair_mla.py
View File

@@ -12,13 +12,7 @@ from vllm.config import VllmConfig, get_current_vllm_config
from vllm.distributed import get_tensor_model_parallel_world_size
from vllm.model_executor.layers.linear import (LinearBase,
UnquantizedLinearMethod)
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import cdiv, round_down
else:
from vllm.utils.math_utils import cdiv, round_down
from vllm.utils.math_utils import cdiv, round_down
import vllm_ascend.envs as envs_ascend
from vllm_ascend.ascend_config import get_ascend_config

9
vllm_ascend/torchair/torchair_mtp_proposer.py
View File

@@ -11,6 +11,7 @@ from vllm.model_executor.layers.attention_layer_base import AttentionLayerBase
from vllm.model_executor.model_loader import get_model_loader
from vllm.model_executor.model_loader.utils import \
process_weights_after_loading
from vllm.utils.torch_utils import set_default_torch_dtype
from vllm.v1.core.sched.output import SchedulerOutput
from vllm.v1.sample.metadata import SamplingMetadata
from vllm.v1.spec_decode.metadata import SpecDecodeMetadata
@@ -23,13 +24,7 @@ from vllm_ascend.torchair.models.torchair_deepseek_mtp import \
TorchairDeepSeekMTP
from vllm_ascend.torchair.utils import (TORCHAIR_CACHE_DIR,
TorchairCommonAttentionMetadata)
from vllm_ascend.utils import (ProfileExecuteDuration, lmhead_tp_enable,
vllm_version_is)
if vllm_version_is("0.11.0"):
from vllm.model_executor.model_loader.utils import set_default_torch_dtype
else:
from vllm.utils.torch_utils import set_default_torch_dtype
from vllm_ascend.utils import ProfileExecuteDuration, lmhead_tp_enable
PADDING_SLOT_ID = -1

8
vllm_ascend/torchair/torchair_sfa.py
View File

@@ -12,13 +12,7 @@ from vllm.config import VllmConfig, get_current_vllm_config
from vllm.distributed import get_tensor_model_parallel_world_size, get_tp_group
from vllm.model_executor.layers.linear import (LinearBase,
UnquantizedLinearMethod)
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import cdiv, round_down
else:
from vllm.utils.math_utils import cdiv, round_down
from vllm.utils.math_utils import cdiv, round_down
import vllm_ascend.envs as envs_ascend
from vllm_ascend.ascend_config import get_ascend_config