Triton Inference Server Release 24.07
The Triton Inference Server container image, release 24.07, is available on NGC and is open source on GitHub.
Contents of the Triton Inference Server container
The Triton Inference Server Docker image contains the inference server executable and related shared libraries in /opt/tritonserver.
For a complete list of what the container includes, refer to Deep Learning Frameworks Support Matrix.
- Ubuntu 22.04 including Python 3.10
- NVIDIA CUDA 12.5.1
- NVIDIA cuBLAS 12.5.3.2
- cuDNN 9.2.1.18
- NVIDIA NCCL 2.22.3 (optimized for NVIDIA NVLink®)
- NVIDIA TensorRT™ 10.2.0.19
- OpenUCX 1.15.0
- GDRCopy 2.3
- NVIDIA HPC-X 2.19
- OpenMPI 4.1.7
- FIL
- NVIDIA DALI® 1.39
- nvImageCodec 0.2.0.7
- ONNX Runtime 1.18.1
- Intel OpenVINO 2024.0.0
- DCGM 3.2.6
- TensorRT-LLM version release/0.11
- vLLM version 0.5.0 post 1
Driver Requirements
Release 24.07 is based on CUDA 12.5.1 which requires NVIDIA Driver release 555 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 470.57 (or later R470), 525.85 (or later R525), 535.86 (or later R535), or 545.23 (or later R545).
The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, R450, R460, R510, R520 and R545 drivers, which are not forward-compatible with CUDA 12.5. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.
GPU Requirements
Release 24.07 supports CUDA compute capability 6.0 and later. This corresponds to GPUs in the NVIDIA Pascal, NVIDIA Volta™, NVIDIA Turing™, NVIDIA Ampere architecture, NVIDIA Hopper™, and NVIDIA Ada Lovelace architecture families. For a list of GPUs to which this compute capability corresponds, see CUDA GPUs. For additional support details, see Deep Learning Frameworks Support Matrix.
Key Features and Enhancements
This Inference Server release includes the following key features and enhancements.
- OpenAI-compatible embeddings and Hugging Face TEI re-ranker API-compatible rankings can now be profiled via GenAI-Perf.
- GenAI-Perf can now receive multiple user-specified prompts via --input-file.
- The request-rate for async requests have been updated in
the OpenAI and HTTP clients to send requests at
exactly that rate. Users submitting more requests
than their models can handle can see increased
latency.
- The stabilization metric for Perf Analyzer has been updated due to these changes, so if latency does not stabilize for async models, a warning will be printed but Perf Analyzer will still complete.
- Perf Analyzer will not validate any user-supplied inputs and outputs, returning an error if the model does not contain them.
- Triton now supports tracing custom activities in the backend. For more information please refer to documentation.
- Enhanced Failure Count Metrics to reflect failure reason of inference request.
NVIDIA Triton Inference Server Container Versions
The following table shows what versions of Ubuntu, CUDA, Triton Inference Server, and NVIDIA TensorRT™ are supported in each of the NVIDIA containers for Triton Inference Server. For older container versions, refer to the Frameworks Support Matrix.
Known Issues
- When using Python models indecoupled mode, users need to ensure that the ResponseSender goes out of scope or is properly cleaned up before unloading the model to guarantee that the unloading process executes correctly.
- Restart support was temporarily removed for Python models.
- Triton TensorRT-LLM Backend container image uses TensorRT-LLM version 0.11.0 and built out of nvcr.io/nvidia/pytorch:24.05-py3.
- The Triton Inference Server with vLLM backend (vllm
version v0.5.0.post1),when using explicit model
control mode, does not support running vLLM models
with the default
distributed_executor_backend and
tensor parallelism sizes greater than 1. Attempting
to load a vLLM model in explicit mode with tensor
parallelism >1 will result in failure at the
initialize step: could not acquire lock for
<_io.BufferedWriter name=<stdout>>
at interpreter shutdown, possibly due to daemon
threads. Please specify
distributed_executor_backend:ray
in the model.json when deploying
vllm models with tensor parallelism > 1.
The Triton Inference Server with vLLM backend currently does not support running vLLM models with tensor parallelism sizes greater than 1 and default "distributed_executor_backend" setting when using explicit model control mode. In attempt to load a vllm model (tp > 1) in explicit mode, users expect to see failure at the `initialize` step: `could not acquire lock for <_io.BufferedWriter name='<stdout>'> at interpreter shutdown, possibly due to daemon threads`. Please specify distributed_executor_backend:ray in the model.json when deploying vllm models with tensor parallelism > 1.
- When loading models with file override, multiple model configuration files are not supported. Users must provide the model configuration by setting parameter config : <JSON> instead of custom configuration file in the following format: file:configs/<model-config-name>.pbtxt : <base64-encoded-file-content>.
- Perf Analyzer no longer supports the --trace-file option.
- TensorRT-LLM backend provides limited support of Triton extensions and features.
- The TensorRT-LLM backend may core dump on server shutdown. This impacts server teardown only and will not impact inferencing.
- The Java CAPI is known to have intermittent segfaults.
- Some systems which implement malloc() may not release memory back to the operating system right away causing a false memory leak. This can be mitigated by using a different malloc implementation. Tcmalloc and jemalloc are installed in the Triton container and can be used by specifying the library in LD_PRELOAD. NVIDIA recommends experimenting with both tcmalloc and jemalloc to determine which one works better for your use case.
- Auto-complete may cause an increase in server start time. To avoid a start time increase, users can provide the full model configuration and launch the server with --disable-auto-complete-config.
- Auto-complete does not support PyTorch models due to lack of metadata in the model. It can only verify that the number of inputs and the input names matches what is specified in the model configuration. There is no model metadata about the number of outputs and datatypes. Related PyTorch bug:https://github.com/pytorch/pytorch/issues/38273.
- Triton Client PIP wheels for ARM SBSA are not available from PyPI and pip will install an incorrect Jetson version of Triton Client library for Arm SBSA. The correct client wheel file can be pulled directly from the Arm SBSA SDK image and manually installed.
- Traced models in PyTorch seem to create overflows when int8 tensor values are transformed to int32 on the GPU. Refer to pytorch/pytorch#66930 for more information.
- Triton cannot retrieve GPU metrics with MIG-enabled GPU devices.
- Triton metrics might not work if the host machine is running a separate DCGM agent on bare-metal or in a container.
- When cloud storage (AWS, GCS, AZURE) is used as a model repository and a model has multiple versions, Triton creates an extra local copy of the cloud model’s folder in the temporary directory, which is deleted upon server’s shutdown.
- Python backend support for Windows is limited and does
not currently support the following features:
- GPU tensors
- CPU and GPU-related metrics
- Custom execution environments
- The model load/unload APIs
- Starting in 24.06, if you use Triton's iGPU container
you might encounter this error message when loading
TensorRT models built with the 24.06 TensorRT iGPU
container:
"Serialization (Serialization assertion stdVersionRead == serializationVersion failed.Version tag does not match. Note: Current Version: 236, Serialized Engine Version: 237)."
If this happens you can rebuild your iGPU models with the 24.04 TensorRT iGPU container and then run them in the Triton 24.06 iGPU container.