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 16, 15:45 UTC
[Bugfix] fix docker build for xpu (#8652)
Yan Ma committed
Sep 23, 2024 at 13:54 UTC
d23679eb9960ad2a876b88ebd0028dbe55c3172a
3 files changed
+6
-14
Dockerfile.xpu
+1
-9
@@ -10,19 +10,11 @@ RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRO
10
RUN apt-get update -y \
11
&& apt-get install -y curl libicu70 lsb-release git wget vim numactl python3 python3-pip ffmpeg libsm6 libxext6 libgl1
12
13
-RUN git clone https://github.com/intel/pti-gpu && \
14
- cd pti-gpu/sdk && \
15
- mkdir build && \
16
- cd build && \
17
- cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_TOOLCHAIN_FILE=../cmake/toolchains/icpx_toolchain.cmake -DBUILD_TESTING=OFF .. && \
18
- make -j && \
19
- cmake --install . --config Release --prefix "/usr/local"
20
-
13
COPY ./ /workspace/vllm
14
15
WORKDIR /workspace/vllm
16
25
-RUN pip install -v -r requirements-xpu.txt
17
+RUN pip install -v -r requirements-xpu.txt --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
18
19
RUN VLLM_TARGET_DEVICE=xpu python3 setup.py install
20
docs/source/getting_started/xpu-installation.rst
+3
-3
@@ -17,8 +17,8 @@ Requirements
17
------------
18
19
* OS: Linux
20
-* Supported Hardware: Intel Data Center GPU (Intel ARC GPU WIP)
21
-* OneAPI requirements: oneAPI 2024.1
20
+* Supported Hardware: Intel Data Center GPU, Intel ARC GPU
21
+* OneAPI requirements: oneAPI 2024.2
22
23
.. _xpu_backend_quick_start_dockerfile:
24
@@ -40,7 +40,7 @@ Quick start using Dockerfile
40
Build from source
41
-----------------
42
43
-- First, install required driver and intel OneAPI 2024.1 or later.
43
+- First, install required driver and intel OneAPI 2024.2 or later.
44
45
- Second, install Python packages for vLLM XPU backend building:
46
requirements-xpu.txt
+2
-2
@@ -3,10 +3,10 @@
3
4
setuptools < 70.0.0 # IPEX's torch have some dependency. to be removed.
5
6
+ray >= 2.9
7
+# Following pkgs retrieved from https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
8
torch == 2.3.1+cxx11.abi
9
intel-extension-for-pytorch == 2.3.110+xpu
10
oneccl_bind_pt == 2.3.100+xpu
11
12
triton-xpu == 3.0.0b2
11
-
12
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/