minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[AMD]: Enable NIXL PD disaggregation for ROCm(1/n) (#28348)

Browse Source 
Co-authored-by: bingxche <bingxche@amd.com>
This commit is contained in:
Zhaoyi Li
2026-06-29 21:23:09 -07:00
committed by GitHub
co-authored by bingxche
parent 2068ae7eec
commit 6bf15aa2b9
6 changed files with 295 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/pr-test-amd-rocm720.yml
+17
View File
@@ -1085,6 +1085,23 @@ jobs:
- name: Install dependencies
run: bash scripts/ci/amd/amd_ci_install_dependency.sh
- name: Verify NIXL in Container
run: |
docker exec ci_sglang bash -lc '
set -eux
export LD_LIBRARY_PATH=/opt/ucx/lib:/opt/rocm/lib:${LD_LIBRARY_PATH:-}
python -c "import nixl; print(\"nixl OK:\", nixl.__file__)"
test -d /opt/ucx/lib
UCX_INFO=$(command -v ucx_info || true)
if [ -z "$UCX_INFO" ]; then
UCX_INFO=/opt/ucx/bin/ucx_info
fi
test -x "$UCX_INFO"
"$UCX_INFO" -v
"$UCX_INFO" -d | grep -iq rocm
python -c "from nixl._api import nixl_agent, nixl_agent_config; nixl_agent(\"ci-smoke\", nixl_agent_config(backends=[\"UCX\"])); print(\"nixl UCX backend init OK\")"
'
- name: Verify RDMA in Container
run: |
docker exec -u root ci_sglang bash -c '
.github/workflows/pr-test-amd.yml
+17
View File
@@ -1134,6 +1134,23 @@ jobs:
- name: Install dependencies
run: bash scripts/ci/amd/amd_ci_install_dependency.sh
- name: Verify NIXL in Container
run: |
docker exec ci_sglang bash -lc '
set -eux
export LD_LIBRARY_PATH=/opt/ucx/lib:/opt/rocm/lib:${LD_LIBRARY_PATH:-}
python -c "import nixl; print(\"nixl OK:\", nixl.__file__)"
test -d /opt/ucx/lib
UCX_INFO=$(command -v ucx_info || true)
if [ -z "$UCX_INFO" ]; then
UCX_INFO=/opt/ucx/bin/ucx_info
fi
test -x "$UCX_INFO"
"$UCX_INFO" -v
"$UCX_INFO" -d | grep -iq rocm
python -c "from nixl._api import nixl_agent, nixl_agent_config; nixl_agent(\"ci-smoke\", nixl_agent_config(backends=[\"UCX\"])); print(\"nixl UCX backend init OK\")"
'
- name: Verify RDMA in Container
run: |
docker exec -u root ci_sglang bash -c '
.github/workflows/release-docker-amd-nightly.yml
+1 -1
View File
@@ -87,7 +87,7 @@ jobs:
# remove --build-arg NIC_BACKEND=ainic for auto detection nic support in mori
# UBUNTU_MIRROR forces apt over HTTPS to dodge port-80 reachability flakes
# to Canonical's archive.ubuntu.com mirror IPs from the amd-docker-scale runner.
docker build . -f docker/rocm.Dockerfile --build-arg SGL_BRANCH=${{ github.ref_name }} --build-arg BUILD_TYPE=${{ matrix.build_type }} --build-arg GPU_ARCH=${{ matrix.gpu_arch }} --build-arg ENABLE_MORI=1 --build-arg SETUPTOOLS_SCM_PRETEND_VERSION=${pretend_version} --build-arg UBUNTU_MIRROR=https://archive.ubuntu.com -t rocm/sgl-dev:${tag}-${{ env.DATE }} --no-cache
docker build . -f docker/rocm.Dockerfile --build-arg SGL_BRANCH=${{ github.ref_name }} --build-arg BUILD_TYPE=${{ matrix.build_type }} --build-arg GPU_ARCH=${{ matrix.gpu_arch }} --build-arg ENABLE_MORI=1 --build-arg ENABLE_NIXL=1 --build-arg SETUPTOOLS_SCM_PRETEND_VERSION=${pretend_version} --build-arg UBUNTU_MIRROR=https://archive.ubuntu.com -t rocm/sgl-dev:${tag}-${{ env.DATE }} --no-cache
docker push rocm/sgl-dev:${tag}-${{ env.DATE }}
# Persist the tag right after rocm/sgl-dev push succeeds so the local
.github/workflows/release-docker-amd-rocm720-nightly.yml
+1 -1
View File
@@ -96,7 +96,7 @@ jobs:
# remove --build-arg NIC_BACKEND=ainic for auto detection nic support in mori
# UBUNTU_MIRROR forces apt over HTTPS to dodge port-80 reachability flakes
# to Canonical's archive.ubuntu.com mirror IPs from the amd-docker-scale runner.
docker build . -f docker/rocm.Dockerfile --build-arg SGL_BRANCH=${{ github.ref_name }} --build-arg BUILD_TYPE=${{ matrix.build_type }} --build-arg GPU_ARCH=${{ matrix.gpu_arch }} --build-arg ENABLE_MORI=1 --build-arg SETUPTOOLS_SCM_PRETEND_VERSION=${pretend_version} --build-arg UBUNTU_MIRROR=https://archive.ubuntu.com -t rocm/sgl-dev:${tag}-${{ env.DATE }} --no-cache
docker build . -f docker/rocm.Dockerfile --build-arg SGL_BRANCH=${{ github.ref_name }} --build-arg BUILD_TYPE=${{ matrix.build_type }} --build-arg GPU_ARCH=${{ matrix.gpu_arch }} --build-arg ENABLE_MORI=1 --build-arg ENABLE_NIXL=1 --build-arg SETUPTOOLS_SCM_PRETEND_VERSION=${pretend_version} --build-arg UBUNTU_MIRROR=https://archive.ubuntu.com -t rocm/sgl-dev:${tag}-${{ env.DATE }} --no-cache
docker push rocm/sgl-dev:${tag}-${{ env.DATE }}
# Persist the tag right after rocm/sgl-dev push succeeds so the local