[Feature] Prefill Context Parallel (PCP) basic support (#28718)

Signed-off-by: QiuChunshuo <qiuchunshuo@huawei.com>
Signed-off-by: FENP <yuanyongjie.yyj@antgroup.com>
Signed-off-by: LookAround <lixushi@huawei.com>
Signed-off-by: Jingchun Gao <gaojingchun1@huawei.com>
Signed-off-by: zhenwenqi2024 <zhenwenqi_2022@qq.com>
Co-authored-by: FENP <yuanyongjie.yyj@antgroup.com>
Co-authored-by: LookAround <lixushi@huawei.com>
Co-authored-by: Jingchun Gao <gaojingchun1@huawei.com>
Co-authored-by: zhenwenqi2024 <zhenwenqi_2022@qq.com>
Co-authored-by: Jingchun Gao <63247409+gjc0824@users.noreply.github.com>
This commit is contained in:
Qiu
2025-11-19 15:52:44 -05:00
committed by GitHub
co-authored by FENP LookAround Jingchun Gao zhenwenqi2024 Jingchun Gao
parent 02f5903b84
commit 2fd893b4ce
27 changed files with 399 additions and 114 deletions
@@ -15,7 +15,11 @@ from tests.kernels.quantization.nvfp4_utils import (
)
from tests.kernels.utils import torch_experts
from vllm.config import VllmConfig
from vllm.distributed import get_dp_group, get_tensor_model_parallel_world_size
from vllm.distributed import (
get_dp_group,
get_pcp_group,
get_tensor_model_parallel_world_size,
)
from vllm.forward_context import set_forward_context
from vllm.model_executor.layers.fused_moe.config import (
FusedMoEConfig,
@@ -561,6 +565,7 @@ def make_modular_kernel(
# make moe config
moe_parallel_config: FusedMoEParallelConfig = FusedMoEParallelConfig.make(
tp_size_=get_tensor_model_parallel_world_size(),
pcp_size_=get_pcp_group().world_size,
dp_size_=get_dp_group().world_size,
vllm_parallel_config=vllm_config.parallel_config,
)