TensorRT-ExecutionProvider.md

TensortRT Execution Provider

The TensorRT execution provider in the ONNX Runtime makes use of NVIDIA's TensortRT Deep Learning inferencing engine to accelerate ONNX model in their family of GPUs. Microsoft and NVIDIA worked closely to integrate the TensorRT execution provider with ONNX Runtime.

With the TensorRT execution provider, the ONNX Runtime delivers better inferencing performance on the same hardware compared to generic GPU acceleration.

Build

For build instructions, please see the BUILD page.

The TensorRT execution provider for ONNX Runtime is built and tested with TensorRT 7.1.3.4.

Using the TensorRT execution provider

C/C++

The TensorRT execution provider needs to be registered with ONNX Runtime to enable in the inference session.

string log_id = "Foo";
auto logging_manager = std::make_unique<LoggingManager>
(std::unique_ptr<ISink>{new CLogSink{}},
                                  static_cast<Severity>(lm_info.default_warning_level),
                                  false,
                                  LoggingManager::InstanceType::Default,
                                  &log_id)
Environment::Create(std::move(logging_manager), env)
InferenceSession session_object{so,env};
session_object.RegisterExecutionProvider(std::make_unique<::onnxruntime::TensorrtExecutionProvider>());
status = session_object.Load(model_file_name);

The C API details are here.

Shape Inference for TensorRT Subgraphs

If some operators in the model are not supported by TensorRT, ONNX Runtime will partition the graph and only send supported subgraphs to TensorRT execution provider. Because TensorRT requires that all inputs of the subgraphs have shape specified, ONNX Runtime will throw error if there is no input shape info. In this case please run shape inference for the entire model first by running script here.

Sample

This example shows how to run Faster R-CNN model on TensorRT execution provider,

First, download Faster R-CNN onnx model from onnx model zoo here.

Second, infer shapes in the model by running shape inference script here,

python symbolic_shape_infer.py --input /path/to/onnx/model/model.onnx --output /path/to/onnx/model/new_model.onnx --auto_merge

Third, replace original model with the new model and run onnx_test_runner tool under ONNX Runtime build directory,

./onnx_test_runner -e tensorrt /path/to/onnx/model/

Python

When using the Python wheel from the ONNX Runtime build with TensorRT execution provider, it will be automatically prioritized over the default GPU or CPU execution providers. There is no need to separately register the execution provider. Python APIs details are .

Sample

Please see this Notebook for an example of running a model on GPU using ONNX Runtime through Azure Machine Learning Services.

Performance Tuning

For performance tuning, please see guidance on this page: ONNX Runtime Perf Tuning

When/if using onnxruntime_perf_test, use the flag -e tensorrt

Configuring environment variables

There are several environment variables for TensorRT execution provider.

• ORT_TENSORRT_MAX_WORKSPACE_SIZE: maximum workspace size for TensorRT engine.

• ORT_TENSORRT_MAX_PARTITION_ITERATIONS: maximum number of iterations allowed in model partitioning for TensorRT. If target model can't be successfully partitioned when the maximum number of iterations is reached, the whole model will fall back to other execution providers such as CUDA or CPU.

• ORT_TENSORRT_MIN_SUBGRAPH_SIZE: minimum node size in a subgraph after partitioning. Subgraphs with smaller size will fall back to other execution providers.

• ORT_TENSORRT_FP16_ENABLE: Enable FP16 mode in TensorRT

• ORT_TENSORRT_ENGINE_CACHE_ENABLE: Enable TensorRT engine caching. The purpose of using engine caching is to save engine build time in the cases that TensorRT may take long time to optimize and build engine. Engine will be cached after it's built at the first time so that next time when inference session is created the engine can be loaded directly from cache. Note each engine is created for specific settings such as precision (FP32/FP16/INT8 etc), workspace, profiles etc, and specific GPUs and it's not portable, so it's essential to make sure those settings are not changing, otherwise the engines need to be rebuilt and cached again. Warning: Please clean up any old engine cache files (.engine) if any of the following changes:

  • Model changes (if there are any changes to the model topology, opset version etc.)
  • ORT version changes (i.e. moving from ORT version 1.4 to 1.5)
  • TensorRT version changes (i.e. moving from TensorRT 7.0 to 7.1)
  • Hardware changes. (Engine files are not portable and optimized for specific Nvidia hardware)

• ORT_TENSORRT_ENGINE_CACHE_PATH: Specify path for TensorRT engine files if ORT_TENSORRT_ENGINE_CACHE_ENABLE is 1

By default TensorRT execution provider builds an ICudaEngine with max workspace size = 1 GB, max partition iterations = 1000, min subgraph size = 1, FP16 mode is disabled and TensorRT engine caching is disabled.

One can override these defaults by setting environment variables ORT_TENSORRT_MAX_WORKSPACE_SIZE, ORT_TENSORRT_MAX_PARTITION_ITERATIONS, ORT_TENSORRT_MIN_SUBGRAPH_SIZE, ORT_TENSORRT_FP16_ENABLE, ORT_TENSORRT_ENGINE_CACHE_ENABLE and ORT_TENSORRT_ENGINE_CACHE_PATH. e.g. on Linux

override default max workspace size to 2GB

export ORT_TENSORRT_MAX_WORKSPACE_SIZE=2147483648

override default maximum number of iterations to 10

export ORT_TENSORRT_MAX_PARTITION_ITERATIONS=10

override default minimum subgraph node size to 5

export ORT_TENSORRT_MIN_SUBGRAPH_SIZE=5

Enable FP16 mode in TensorRT

export ORT_TENSORRT_FP16_ENABLE=1

Enable TensorRT engine caching

export ORT_TENSORRT_ENGINE_CACHE_ENABLE=1

• Please Note warning above. This feature is experimental. Engine cache files must be invalidated if there are any changes to the model, ORT version, TensorRT version or if the underlying hardware changes. Engine files are not portable across devices.

Specify TensorRT engine cache path

export ORT_TENSORRT_ENGINE_CACHE_PATH="/path/to/cache"