chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

[NPU] upgrade npu mf_apater plugin (#15853)

Browse Source
This commit is contained in:
James
2026-01-13 09:02:10 +08:00
committed by GitHub
parent 1f0e3d7fd8
commit ae0baefb94
6 changed files with 10 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docker/npu.Dockerfile
View File

@@ -57,7 +57,7 @@ ENV LC_ALL=en_US.UTF-8
### Install MemFabric
RUN ${PIP_INSTALL} mf-adapter==1.0.0
RUN ${PIP_INSTALL} memfabric-hybrid==1.0.0
### Install SGLang Model Gateway
RUN ${PIP_INSTALL} sglang-router

4
docs/advanced_features/pd_disaggregation.md
View File

@@ -266,10 +266,10 @@ python -m sglang.launch_server \
### Usage
Use ascend backend with [mf_adapter(download link)](https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com:443/sglang/mf_adapter-1.0.0-cp311-cp311-linux_aarch64.whl?AccessKeyId=HPUAXT4YM0U8JNTERLST&Expires=1783151861&Signature=3j10QDUjqk70enaq8lostYV2bEA%3D) and ASCEND_MF_STORE_URL being set
Use ascend backend with [memfabric_hybrid](https://gitcode.com/Ascend/memfabric_hybrid) and ASCEND_MF_STORE_URL being set
```bash
pip install mf_adapter-1.0.0-cp311-cp311-linux_aarch64.whl --force-reinstall
pip install memfabric-hybrid==1.0.0
export ASCEND_MF_STORE_URL="tcp://xxx.xx.xxx.xxx:xxxx"
```
Use mooncake backend, more details can be found in mooncake section.

6
docs/platforms/ascend_npu.md
View File

@@ -41,12 +41,12 @@ conda activate sglang_npu
Prior to start work with SGLang on Ascend you need to install CANN Toolkit, Kernels operator package and NNAL version 8.3.RC2 or higher, check the [installation guide](https://www.hiascend.com/document/detail/zh/CANNCommunityEdition/83RC1/softwareinst/instg/instg_0008.html?Mode=PmIns&InstallType=local&OS=openEuler&Software=cannToolKit)
#### MemFabric Adaptor
#### MemFabric-Hybrid
If you want to use PD disaggregation mode, you need to install MemFabric Adaptor. MemFabric Adaptor is a drop-in replacement of Mooncake Transfer Engine that enables KV cache transfer on Ascend NPU clusters.
If you want to use PD disaggregation mode, you need to install MemFabric-Hybrid. MemFabric-Hybrid is a drop-in replacement of Mooncake Transfer Engine that enables KV cache transfer on Ascend NPU clusters.
```shell
pip install mf-adapter==1.0.0
pip install memfabric-hybrid==1.0.0
```
#### Pytorch and Pytorch Framework Adaptor on Ascend

4
python/sglang/srt/disaggregation/ascend/transfer_engine.py
View File

@@ -8,7 +8,7 @@ from sglang.srt.disaggregation.mooncake.transfer_engine import MooncakeTransferE
from sglang.srt.disaggregation.utils import DisaggregationMode
try:
from mf_adapter import TransferEngine
from memfabric_hybrid import TransferEngine
import_error = None
except ImportError as e:
@@ -25,7 +25,7 @@ class AscendTransferEngine(MooncakeTransferEngine):
):
if import_error is not None:
logger.warning(
"Please install mf_adapter, for details, see docs/backend/pd_disaggregation.md"
"Please install memfabric_hybrid, for details, see docs/backend/pd_disaggregation.md"
)
raise import_error

2
python/sglang/srt/managers/disagg_service.py
View File

@@ -34,7 +34,7 @@ def start_disagg_service(
)
if is_create_store:
try:
from mf_adapter import create_config_store
from memfabric_hybrid import create_config_store
ascend_url = os.getenv("ASCEND_MF_STORE_URL")
create_config_store(ascend_url)

2
scripts/ci/npu_ci_install_dependency.sh
View File

@@ -25,7 +25,7 @@ ${PIP_INSTALL} wheel==0.45.1 pybind11
### Install MemFabric
${PIP_INSTALL} mf-adapter==1.0.0
${PIP_INSTALL} memfabric-hybrid==1.0.0
### Install PyTorch and PTA