xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-23 07:45:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Build/CI] Enabling AMD Entrypoints Test (#4834)

Browse Source 
Co-authored-by: Alexey Kondratiev <alexey.kondratiev@amd.com>
This commit is contained in:
Alexei-V-Ivanov-AMD 2024-05-20 13:29:28 -05:00 committed by GitHub
parent 546a97ef69
commit 943e72ca56
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 16 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.buildkite/test-pipeline.yaml
View File

@ -60,7 +60,8 @@ steps:
command: pytest -v -s engine tokenization test_sequence.py test_config.py test_logger.py command: pytest -v -s engine tokenization test_sequence.py test_config.py test_logger.py
- label: Entrypoints Test - label: Entrypoints Test
#mirror_hardwares: [amd] mirror_hardwares: [amd]
commands: commands:
# these tests have to be separated, because each one will allocate all posible GPU memory # these tests have to be separated, because each one will allocate all posible GPU memory
- pytest -v -s entrypoints --ignore=entrypoints/test_server_oot_registration.py - pytest -v -s entrypoints --ignore=entrypoints/test_server_oot_registration.py

8
Dockerfile.rocm
View File

@ -92,19 +92,23 @@ RUN if [ "$BUILD_TRITON" = "1" ]; then \
WORKDIR /vllm-workspace WORKDIR /vllm-workspace
COPY . . COPY . .
#RUN python3 -m pip install pynvml # to be removed eventually
RUN python3 -m pip install --upgrade pip numba RUN python3 -m pip install --upgrade pip numba
# make sure punica kernels are built (for LoRA) # make sure punica kernels are built (for LoRA)
ENV VLLM_INSTALL_PUNICA_KERNELS=1 ENV VLLM_INSTALL_PUNICA_KERNELS=1
# Workaround for ray >= 2.10.0
ENV RAY_EXPERIMENTAL_NOSET_ROCR_VISIBLE_DEVICES=1
ENV VLLM_NCCL_SO_PATH=/opt/rocm/lib/librccl.so
RUN --mount=type=cache,target=/root/.cache/pip \ RUN --mount=type=cache,target=/root/.cache/pip \
pip install -U -r requirements-rocm.txt \ pip install -U -r requirements-rocm.txt \
&& patch /opt/rocm/include/hip/amd_detail/amd_hip_bf16.h ./rocm_patch/rocm_bf16.patch \ && patch /opt/rocm/include/hip/amd_detail/amd_hip_bf16.h ./rocm_patch/rocm_bf16.patch \
&& python3 setup.py install \ && python3 setup.py install \
&& cp build/lib.linux-x86_64-cpython-39/vllm/_C.cpython-39-x86_64-linux-gnu.so vllm/ \ && cp build/lib.linux-x86_64-cpython-39/vllm/_C.cpython-39-x86_64-linux-gnu.so vllm/ \
&& cp build/lib.linux-x86_64-cpython-39/vllm/_punica_C.cpython-39-x86_64-linux-gnu.so vllm/ \
&& cd .. && cd ..
RUN python3 -m pip install --upgrade pip
RUN python3 -m pip install --no-cache-dir ray[all]==2.9.3
CMD ["/bin/bash"] CMD ["/bin/bash"]

3
requirements-rocm.txt
View File

@ -2,4 +2,5 @@
-r requirements-common.txt -r requirements-common.txt
# Dependencies for AMD GPUs # Dependencies for AMD GPUs
ray == 2.9.3 ray >= 2.10.0
pytest-asyncio

8
tests/spec_decode/e2e/conftest.py
View File

@ -6,8 +6,12 @@ from typing import Dict, List, Optional, Tuple, Union
import pytest import pytest
import ray import ray
import torch import torch
from pynvml import (nvmlDeviceGetHandleByIndex, nvmlDeviceGetMemoryInfo,
nvmlInit) from vllm.utils import is_hip
if (not is_hip()):
from pynvml import (nvmlDeviceGetHandleByIndex, nvmlDeviceGetMemoryInfo,
nvmlInit)
from vllm import LLM from vllm import LLM
from vllm.engine.arg_utils import AsyncEngineArgs from vllm.engine.arg_utils import AsyncEngineArgs