Skip to content

[TRTLLM-9247][infra] Upgrade NIXL to 0.7.1 #9055

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
bo-nv merged 9 commits into from
Nov 20, 2025
Merged

[TRTLLM-9247][infra] Upgrade NIXL to 0.7.1 #9055

Show file tree
Hide file tree
Changes from 8 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
9e4f6c5
[TRTLLM-9247][infra] Upgrade NIXL to 0.7.1
bo-nv Nov 11, 2025
8b3da14
Update docker images
bo-nv Nov 12, 2025
f496684
back to ucx 1.19
bo-nv Nov 14, 2025
86de216
update docker images
bo-nv Nov 14, 2025
0f6f8a8
update docker images
bo-nv Nov 15, 2025
a07f410
fix test
jthomson04 Nov 18, 2025
e315e26
Merge pull request #1 from jthomson04/jthomson04/fix-disagg-connector…
bo-nv Nov 18, 2025
ec2f2aa
Update jenkins/current_image_tags.properties
bo-nv Nov 20, 2025
920e79d
Update jenkins/current_image_tags.properties
bo-nv Nov 20, 2025
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
2 changes: 1 addition & 1 deletion docker/common/install_nixl.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ set -ex
GITHUB_URL="https://github.com"
UCX_INSTALL_PATH="/usr/local/ucx/"
CUDA_PATH="/usr/local/cuda"
NIXL_VERSION="0.5.0"
NIXL_VERSION="0.7.1"
NIXL_REPO="https://github.com/ai-dynamo/nixl.git"
OLD_LD_LIBRARY_PATH=$LD_LIBRARY_PATH

Expand Down
8 changes: 4 additions & 4 deletions jenkins/current_image_tags.properties
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@
# images are adopted from PostMerge pipelines, the abbreviated commit hash is used instead.
IMAGE_NAME=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm

LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.10-py3-x86_64-ubuntu24.04-trt10.13.3.9-skip-tritondevel-202511131803-8929
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.10-py3-aarch64-ubuntu24.04-trt10.13.3.9-skip-tritondevel-202511131803-8929
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-13.0.2-devel-rocky8-x86_64-rocky8-py310-trt10.13.3.9-skip-tritondevel-202511131803-8929
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-13.0.2-devel-rocky8-x86_64-rocky8-py312-trt10.13.3.9-skip-tritondevel-202511131803-8929
LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-tritondevel-torch_skip-86de216-github-tritondevel-202511191426-9055
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:sbsa-tritondevel-torch_skip-86de216-github-tritondevel-202511191426-9055
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-rockylinux8-torch_skip-py310-86de216-github-tritondevel-202511191426-9055
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-rockylinux8-torch_skip-py312-86de216-github-tritondevel-202511191426-9055
21 changes: 16 additions & 5 deletions tests/integration/defs/llmapi/test_llm_api_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -363,11 +363,15 @@ def test_connector_disagg_prefill(enforce_single_worker, model_with_connector,
save_async):
model_fn, scheduler, worker = model_with_connector

model = model_fn(
prefill_worker = model_fn(
disable_overlap_scheduler=True,
cache_transceiver_config=CacheTransceiverConfig(backend="DEFAULT"))

sampling_params = SamplingParams(ignore_eos=True)
decode_worker = model_fn(
cache_transceiver_config=CacheTransceiverConfig(backend="DEFAULT"),
kv_connector_config=None)

sampling_params = SamplingParams(ignore_eos=True, max_tokens=16)

disaggregated_params = DisaggregatedParams(request_type="context_only")

Expand All @@ -382,9 +386,16 @@ def test_connector_disagg_prefill(enforce_single_worker, model_with_connector,
scheduler.request_finished.return_value = False
worker.get_finished.return_value = [], []

model.generate([0] * 48,
sampling_params=sampling_params,
disaggregated_params=disaggregated_params)
result = prefill_worker.generate([0] * 48,
sampling_params=sampling_params,
disaggregated_params=disaggregated_params)

gen_disagg_params = result.disaggregated_params
gen_disagg_params.request_type = "generation_only"

result = decode_worker.generate([0] * 48,
sampling_params=sampling_params,
disaggregated_params=gen_disagg_params)

assert scheduler.build_connector_meta.call_count == 1

Expand Down