Skip to content

build: bump UCX to v1.21.0 for v1.4.0 (OPS-7781) #12401

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dagil-nvidia merged 4 commits into from
Jul 30, 2026
Merged

build: bump UCX to v1.21.0 for v1.4.0 (OPS-7781) #12401

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e32d260
build(container): bump UCX to v1.21.0 for v1.4.0 (OPS-7781)
pvijayakrish Jul 29, 2026
bcae0b4
Merge branch 'main' into pvijayakrish/bump-ucx-version
pvijayakrish Jul 29, 2026
2249b7f
Merge branch 'main' into pvijayakrish/bump-ucx-version
pvijayakrish Jul 29, 2026
e6ee8ea
ci: raise dynamo-pipeline build timeout default to 60 min
pvijayakrish Jul 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion .github/workflows/dynamo-pipeline.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,9 +25,10 @@ on:
type: boolean
default: false
build_timeout_minutes:
description: 'Timeout for the image build job. Matches the 60 min the other framework builds get in pr.yaml.'
required: false
type: number
default: 30
default: 60
no_cache:
description: 'Disable BuildKit cache-from/to. Nightly sets true for regression detection.'
required: false
Expand Down
2 changes: 1 addition & 1 deletion container/compliance/native_packages.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,7 @@ packages:
# UCX — Unified Communication X, built from source in wheel_builder and used by
# the NIXL transport. Installed to /usr/local (no dpkg metadata).
- name: UCX
version: v1.20.x
version: v1.21.0
license: BSD-3-Clause
source: https://github.com/openucx/ucx
images:
Expand Down
3 changes: 1 addition & 2 deletions container/context.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,9 +42,8 @@ dynamo:
etcd_version: v3.5.30

nixl_ref: v1.0.1
nixl_ucx_ref: v1.20.x
nixl_ucx_ref: v1.21.0
nixl_gdrcopy_ref: v2.5.2
nixl_ucx_efa_ref: 9d2b88a1f67faf9876f267658bd077b379b8bb76
nixl_libfabric_repo: https://github.com/ofiwg/libfabric.git
nixl_libfabric_ref: v2.4.0
# Keep in sync with upstream NIXL's contrib/Dockerfile.manylinux.
Expand Down
9 changes: 3 additions & 6 deletions container/templates/wheel_builder.Dockerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,9 +77,6 @@ ADD --checksum=sha256:f60e802b6f41350393e34b24793db888a8be514054769bd17e7a6e9c0c
https://github.com/intel/xpumanager/releases/download/v1.3.6/xpu-smi_1.3.6_20260206.143628.1004f6cb.u24.04_amd64.deb \
/tmp/xpu-smi.deb

# Fetch UCX patch
RUN wget --tries=3 --waitretry=5 https://raw.githubusercontent.com/intel/llm-scaler/35a14cbc08d714f460a29b7a7328df5620c8530f/vllm/patches/ai-dynamo-xpu/patches/ucx-v1.12.0.patch -O /tmp/ucx.patch

# Install xpu-smi without explicitly changing the Intel compute runtime stack.
RUN apt-get update && \
apt-get install -y --no-install-recommends /tmp/xpu-smi.deb && \
Expand Down Expand Up @@ -376,9 +373,9 @@ RUN --mount=type=secret,id=aws-web-identity-token,target=/run/secrets/aws-token
git clone https://github.com/openucx/ucx.git && \
cd ucx && \
git checkout $NIXL_UCX_REF && \
if [ "$DEVICE" = "xpu" ]; then \
git apply --ignore-whitespace /tmp/ucx.patch; \
fi && \
# The intel/llm-scaler xe-GDR patch (ucx-v1.12.0.patch) is upstream since
# UCX v1.21.0 (ib_md.c xe srcversion check, ze_copy_md.c HOST bit); restore
# the fetch + git apply for DEVICE=xpu if this ref ever drops below v1.21.0.
Comment thread
pvijayakrish marked this conversation as resolved.
./autogen.sh && \
if [ "$DEVICE" = "xpu" ]; then \
./contrib/configure-release \
Expand Down
Loading