Gahow Wang
445e491123
third_party/vllm/ now tracked in git for direct patch management.
Based on vLLM v0.18.1 release with one patch applied:
vllm/v1/core/sched/scheduler.py:
Replace fatal assert with graceful skip when KV transfer callback
arrives for an already-aborted request during PD disaggregated serving.
Future vLLM modifications should be made directly in third_party/vllm/
and committed normally. The patches/ directory is kept as documentation
of what changed from upstream.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
31 lines
831 B
YAML
31 lines
831 B
YAML
group: Hardware - AMD Build
|
|
steps:
|
|
- label: "AMD: :docker: build image"
|
|
key: image-build-amd
|
|
depends_on: []
|
|
device: amd_cpu
|
|
no_plugin: true
|
|
commands:
|
|
- >
|
|
docker build
|
|
--build-arg max_jobs=16
|
|
--build-arg REMOTE_VLLM=1
|
|
--build-arg ARG_PYTORCH_ROCM_ARCH='gfx90a;gfx942;gfx950'
|
|
--build-arg VLLM_BRANCH=$BUILDKITE_COMMIT
|
|
--tag "rocm/vllm-ci:${BUILDKITE_COMMIT}"
|
|
-f docker/Dockerfile.rocm
|
|
--target test
|
|
--no-cache
|
|
--progress plain .
|
|
- docker push "rocm/vllm-ci:${BUILDKITE_COMMIT}"
|
|
env:
|
|
DOCKER_BUILDKIT: "1"
|
|
retry:
|
|
automatic:
|
|
- exit_status: -1 # Agent was lost
|
|
limit: 1
|
|
- exit_status: -10 # Agent was lost
|
|
limit: 1
|
|
- exit_status: 1 # Machine occasionally fail
|
|
limit: 1
|