NVIDIA
Dynamo Tensorrt-LLM Runtime
Container
NVIDIA
Dynamo Tensorrt-LLM Runtime

The Dynamo TensorRT-LLM runtime image is a containerized build of Dynamo + TensorRT-LLM which serves as the base runtime environment for tensorrt-llm based inference with Dynamo's distributed inference framework.

  • LayerLabelCreated
    sha256:cea9da90fc3a25db057913afe5db2627b86726a69e79c5ef29631cc91cb4d59dCOPY
    /legal /legal
    08/21/2026 2:11 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    NVIDIA_DRIVER_CAPABILITIES=video,compute,utility
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4CMD
    /bin/bash
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENTRYPOINT
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4USER
    dynamo
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    DYNAMO_COMMIT_SHA=db4a0d6ec2627cb1873e9c64b94f4a502fc29444
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ARG
    DYNAMO_COMMIT_SHA=db4a0d6ec2627cb1873e9c64b94f4a502fc29444
    08/21/2026 2:10 AM UTC
    sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1WORKDIR
    /workspace
    08/21/2026 2:10 AM UTC
    sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
    DYNAMO_HOME=/workspace HOME=/home/dynamo VIRTUAL_ENV=/opt/dynamo/venv PATH=/opt/dynamo/venv/bin:/opt/uv/bin:/usr/local/bin/etcd:/usr/local/lib/python3.12/dist-packages/torch_tensorrt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin:/opt/tensorrt/bin IMAGEIO_FFMPEG_EXE=/usr/local/bin/ffmpeg LD_PRELOAD=/opt/dynamo/libstdc++.so.6:/usr/local/lib/python3.12/dist-packages/tensorrt_llm/libs/nixl/libnixl.so NIXL_PLUGIN_DIR=/usr/local/lib/python3.12/dist-packages/tensorrt_llm/libs/nixl/plugins NIXL_VERSION=
    08/21/2026 2:10 AM UTC
    sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1RUN
    /bin/bash -c /usr/bin/python3 -c 'import os, re, sys; D = "/usr/local/lib/python3.12/dist-packages"; W = {"aiohttp": ((3, 14, 3), (4, 0, 0)), "pillow": ((12, 3, 0), (12, 4, 0)), "mistune": ((3, 3, 0), (3, 4, 0)), "tornado": ((6, 5, 6), (6, 6, 0)), "jupyter-server": ((2, 20, 0), (2, 21, 0)), "jupyterlab": ((4, 5, 10), (4, 6, 0)), "gitpython": ((3, 1, 58), (3, 2, 0)), "soupsieve": ((2, 8, 4), (2, 9, 0))}; norm = lambda s: re.sub(r"[-_.]+", "-", s).lower(); tv = lambda s: tuple(int(x) for x in re.findall(r"\d+", s)[:3]); stems = [d[:-10].rsplit("-", 1) for d in os.listdir(D) if d.endswith(".dist-info") and "-" in d[:-10]]; found = {k: [] for k in W}; [found[norm(n)].append(v) for n, v in stems if norm(n) in found]; print("post-overlay system-site dist-info:", found); bad = [(k, v) for k, v in sorted(found.items()) if len(v) != 1 or not (W[k][0] <= tv(v[0]) < W[k][1])]; print("FAILED:", bad) if bad else None; sys.exit(1 if bad else 0)' &&
      /usr/bin/python3 -c 'import glob, os, sys; D = "/usr/local/lib/python3.12/dist-packages"; libs = os.path.join(D, "pillow.libs"); recs = glob.glob(os.path.join(D, "pillow-*.dist-info", "RECORD")); sys.exit(1) if len(recs) != 1 else None; owned = {os.path.basename(l.split(",")[0]) for l in open(recs[0]) if l.startswith("pillow.libs/")}; disk = set(os.listdir(libs)) if os.path.isdir(libs) else set(); stale = sorted(disk - owned); print("pillow.libs owned:", len(owned), "on disk:", len(disk)); print("STALE:", stale) if stale else None; sys.exit(1 if stale else 0)'
    08/21/2026 2:10 AM UTC
    ...