A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[XPU] upgrade torch 2.8 on for XPU (#22300)
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
Kunshang Ji committed
Aug 9, 2025 at 08:03 UTC
81c57f60a2c77d169dbec021bb58a467edf580f6
4 files changed
+15
-24
docker/Dockerfile.xpu
+11
-6
@@ -1,9 +1,12 @@
1
-# oneapi 2025.0.2 docker base image use rolling 2448 package. https://dgpu-docs.intel.com/releases/packages.html?release=Rolling+2448.13&os=Ubuntu+22.04, and we don't need install driver manually.
2
-FROM intel/deep-learning-essentials:2025.0.2-0-devel-ubuntu22.04 AS vllm-base
1
+FROM intel/deep-learning-essentials:2025.1.3-0-devel-ubuntu24.04 AS vllm-base
2
3
RUN rm /etc/apt/sources.list.d/intel-graphics.list
4
6
-RUN apt-get update -y && \
5
+RUN apt clean && apt-get update -y && \
6
+ apt-get install -y software-properties-common && \
7
+ add-apt-repository ppa:deadsnakes/ppa && \
8
+ apt-get install -y python3.10 python3.10-distutils && \
9
+ curl -sS https://bootstrap.pypa.io/get-pip.py | python3.10 && \
10
apt-get install -y --no-install-recommends --fix-missing \
11
curl \
12
ffmpeg \
@@ -14,11 +17,13 @@ RUN apt-get update -y && \
17
libgl1 \
18
lsb-release \
19
numactl \
17
- python3 \
18
- python3-dev \
19
- python3-pip \
20
+ python3.10-dev \
21
wget
22
23
+
24
+RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
25
+RUN update-alternatives --install /usr/bin/python python /usr/bin/python3.10 1
26
+
27
WORKDIR /workspace/vllm
28
COPY requirements/xpu.txt /workspace/vllm/requirements/xpu.txt
29
COPY requirements/common.txt /workspace/vllm/requirements/common.txt
requirements/xpu.txt
+3
-8
@@ -10,15 +10,10 @@ wheel
10
jinja2>=3.1.6
11
datasets # for benchmark scripts
12
numba == 0.60.0 # v0.61 doesn't support Python 3.9. Required for N-gram speculative decoding
13
-
14
-torch==2.7.0+xpu
13
+--extra-index-url=https://download.pytorch.org/whl/xpu
14
+torch==2.8.0+xpu
15
torchaudio
16
torchvision
17
pytorch-triton-xpu
18
---extra-index-url=https://download.pytorch.org/whl/xpu
19
-
20
-# Please refer xpu doc, we need manually install intel-extension-for-pytorch 2.6.10+xpu due to there are some conflict dependencies with torch 2.6.0+xpu
21
-# FIXME: This will be fix in ipex 2.7. just leave this here for awareness.
22
-intel-extension-for-pytorch==2.7.10+xpu
23
-oneccl_bind_pt==2.7.0+xpu
18
--extra-index-url=https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
19
+intel-extension-for-pytorch==2.8.10+xpu
vllm/plugins/__init__.py
-9
@@ -4,8 +4,6 @@
4
import logging
5
from typing import Any, Callable
6
7
-import torch
8
-
7
import vllm.envs as envs
8
9
logger = logging.getLogger(__name__)
@@ -68,13 +66,6 @@ def load_general_plugins():
66
return
67
plugins_loaded = True
68
71
- # some platform-specific configurations
72
- from vllm.platforms import current_platform
73
-
74
- if current_platform.is_xpu():
75
- # see https://github.com/pytorch/pytorch/blob/43c5f59/torch/_dynamo/config.py#L158
76
- torch._dynamo.config.disable = True
77
-
69
plugins = load_plugins_by_group(group=DEFAULT_PLUGINS_GROUP)
70
# general plugins, we only need to execute the loaded functions
71
for func in plugins.values():
vllm/v1/worker/xpu_worker.py
+1
-1
@@ -152,7 +152,7 @@ class XPUWorker(Worker):
152
raise RuntimeError(
153
f"Not support device type: {self.device_config.device}")
154
155
- ENV_CCL_ZE_IPC_EXCHANGE = os.getenv("CCL_ZE_IPC_EXCHANGE", "drmfd")
155
+ ENV_CCL_ZE_IPC_EXCHANGE = os.getenv("CCL_ZE_IPC_EXCHANGE", "pidfd")
156
ENV_CCL_ATL_TRANSPORT = os.getenv("CCL_ATL_TRANSPORT", "ofi")
157
ENV_LOCAL_WORLD_SIZE = os.getenv("LOCAL_WORLD_SIZE",
158
str(self.parallel_config.world_size))