官方网站参考：

NVIDIA - CUDA - onnxruntime

要求
请参考下表，了解ONNX运行时推断包的官方GPU包依赖关系。请注意，ONNX运行时训练与PyTorch CUDA版本保持一致；有关支持的版本，请参阅onnxruntime.ai上的“优化训练”选项卡。
由于Nvidia CUDA次要版本兼容性，使用CUDA 11.8构建的ONNX Runtime与任何CUDA 11.x版本兼容；使用CUDA 12.x构建的ONNX Runtime与任何CUDA 12.x版本兼容。
使用cuDNN 8.x构建的ONNX运行时与cuDNN 9.x不兼容，反之亦然。您可以根据与运行时环境匹配的CUDA和cuDNN主要版本选择包（例如，PyTorch 2.3使用cuDNN 8.x，而PyTorch 2.4或更高版本使用cuDNN 9.x）。

CUDA 12.x

To install CUDA 12 package, please look at Install ORT.

ONNX Runtime

CUDA

cuDNN

Notes

1.18.1

12.x

9.x

cuDNN 9 is required. No Java package.

1.18.0

12.x

8.x

Java package is added.

1.17.x

12.x

8.x

Only C++/C# Nuget and Python packages are released. No Java package.

CUDA 11.x

ONNX Runtime

CUDA

cuDNN

Notes

1.18.x

11.8

8.x

1.17

1.16

1.15

11.8

8.2.4 (Linux)

8.5.0.96 (Windows)

Tested with CUDA versions from 11.6 up to 11.8, and cuDNN from 8.2 up to 8.9

1.14

1.13

11.6

8.2.4 (Linux)

8.5.0.96 (Windows)

libcudart 11.4.43

libcufft 10.5.2.100

libcurand 10.2.5.120

libcublasLt 11.6.5.2

libcublas 11.6.5.2

libcudnn 8.2.4

1.12

1.11

11.4

8.2.4 (Linux)

8.2.2.26 (Windows)

libcudart 11.4.43

libcufft 10.5.2.100

libcurand 10.2.5.120

libcublasLt 11.6.5.2

libcublas 11.6.5.2

libcudnn 8.2.4

1.10

11.4

8.2.4 (Linux)

8.2.2.26 (Windows)

libcudart 11.4.43

libcufft 10.5.2.100

libcurand 10.2.5.120

libcublasLt 11.6.1.51

libcublas 11.6.1.51

libcudnn 8.2.4

1.9

11.4

8.2.4 (Linux)

8.2.2.26 (Windows)

libcudart 11.4.43

libcufft 10.5.2.100

libcurand 10.2.5.120

libcublasLt 11.6.1.51

libcublas 11.6.1.51

libcudnn 8.2.4

1.8

11.0.3

8.0.4 (Linux)

8.0.2.39 (Windows)

libcudart 11.0.221

libcufft 10.2.1.245

libcurand 10.2.1.245

libcublasLt 11.2.0.252

libcublas 11.2.0.252

libcudnn 8.0.4

1.7

11.0.3

8.0.4 (Linux)

8.0.2.39 (Windows)

libcudart 11.0.221

libcufft 10.2.1.245

libcurand 10.2.1.245

libcublasLt 11.2.0.252

libcublas 11.2.0.252

libcudnn 8.0.4

CUDA 10.x

ONNX Runtime

CUDA

cuDNN

Notes

1.5-1.6

10.2

8.0.3

CUDA 11 can be built from source

1.2-1.4

10.1

7.6.5

Requires cublas10-10.2.1.243; cublas 10.1.x will not work

1.0-1.1

10.0

7.6.4

CUDA versions from 9.1 up to 10.1, and cuDNN versions from 7.1 up to 7.4 should also work with Visual Studio 2017

For older versions, please reference the readme and build pages on the release branch.

Build

For build instructions, please see the BUILD page.

Configuration Options

The CUDA Execution Provider supports the following configuration options.

device_id

The device ID.

Default value: 0

user_compute_stream

Defines the compute stream for the inference to run on. It implicitly sets the has_user_compute_stream option. It cannot be set through UpdateCUDAProviderOptions, but rather UpdateCUDAProviderOptionsWithValue. This cannot be used in combination with an external allocator.

Example python usage:

providers = [("CUDAExecutionProvider", {"device_id": torch.cuda.current_device(),
                                        "user_compute_stream": str(torch.cuda.current_stream().cuda_stream)})]
sess_options = ort.SessionOptions()
sess = ort.InferenceSession("my_model.onnx", sess_options=sess_options, providers=providers)

To take advantage of user compute stream, it is recommended to use I/O Binding to bind inputs and outputs to tensors in device.

do_copy_in_default_stream

Whether to do copies in the default stream or use separate streams. The recommended setting is true. If false, there are race conditions and possibly better performance.

Default value: true

use_ep_level_unified_stream

Uses the same CUDA stream for all threads of the CUDA EP. This is implicitly enabled by has_user_compute_stream, enable_cuda_graph or when using an external allocator.

Default value: false

gpu_mem_limit

The size limit of the device memory arena in bytes. This size limit is only for the execution provider’s arena. The total device memory usage may be higher. s: max value of C++ size_t type (effectively unlimited)

Note: Will be over-ridden by contents of default_memory_arena_cfg (if specified)

arena_extend_strategy

The strategy for extending the device memory arena.

Value

Description

kNextPowerOfTwo (0)

subsequent extensions extend by larger amounts (multiplied by powers of two)

kSameAsRequested (1)

extend by the requested amount

Default value: kNextPowerOfTwo

Note: Will be over-ridden by contents of default_memory_arena_cfg (if specified)

cudnn_conv_algo_search

The type of search done for cuDNN convolution algorithms.

Value

Description

EXHAUSTIVE (0)

expensive exhaustive benchmarking using cudnnFindConvolutionForwardAlgorithmEx

HEURISTIC (1)

lightweight heuristic based search using cudnnGetConvolutionForwardAlgorithm_v7

DEFAULT (2)

default algorithm using CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_PRECOMP_GEMM

Default value: EXHAUSTIVE

cudnn_conv_use_max_workspace

Check tuning performance for convolution heavy models for details on what this flag does. This flag is only supported from the V2 version of the provider options struct when used using the C API.(sample below)

Default value: 1, for versions 1.14 and later 0, for previous versions

cudnn_conv1d_pad_to_nc1d

Check convolution input padding in the CUDA EP for details on what this flag does. This flag is only supported from the V2 version of the provider options struct when used using the C API. (sample below)

Default value: 0

enable_cuda_graph

Check using CUDA Graphs in the CUDA EP for details on what this flag does. This flag is only supported from the V2 version of the provider options struct when used using the C API. (sample below)

Default value: 0

enable_skip_layer_norm_strict_mode

Whether to use strict mode in SkipLayerNormalization cuda implementation. The default and recommended setting is false. If enabled, accuracy improvement and performance drop can be expected. This flag is only supported from the V2 version of the provider options struct when used using the C API. (sample below)

Default value: 0

use_tf32

TF32 is a math mode available on NVIDIA GPUs since Ampere. It allows certain float32 matrix multiplications and convolutions to run much faster on tensor cores with TensorFloat-32 reduced precision: float32 inputs are rounded with 10 bits of mantissa and results are accumulated with float32 precision.

Default value: 1

TensorFloat-32 is enabled by default. Starting from ONNX Runtime 1.18, you can use this flag to disable it for an inference session.

Example python usage:

providers = [("CUDAExecutionProvider", {"use_tf32": 0})]
sess_options = ort.SessionOptions()
sess = ort.InferenceSession("my_model.onnx", sess_options=sess_options, providers=providers)

This flag is only supported from the V2 version of the provider options struct when used using the C API. (sample below)

gpu_external_[alloc|free|empty_cache]

gpu_external_* is used to pass external allocators. Example python usage:

from onnxruntime.training.ortmodule.torch_cpp_extensions import torch_gpu_allocator

provider_option_map["gpu_external_alloc"] = str(torch_gpu_allocator.gpu_caching_allocator_raw_alloc_address())
provider_option_map["gpu_external_free"] = str(torch_gpu_allocator.gpu_caching_allocator_raw_delete_address())
provider_option_map["gpu_external_empty_cache"] = str(torch_gpu_allocator.gpu_caching_allocator_empty_cache_address())

Default value: 0

prefer_nhwc

This option is not available in default builds ! One has to compile ONNX Runtime with onnxruntime_USE_CUDA_NHWC_OPS=ON. If this is enabled the EP prefers NHWC operators over NCHW. Needed transforms will be added to the model. As NVIDIA tensor cores can only work on NHWC layout this can increase performance if the model consists of many supported operators and does not need too many new transpose nodes. Wider operator support is planned in the future.

This flag is only supported from the V2 version of the provider options struct when used using the C API. The V2 provider options struct can be created using CreateCUDAProviderOptions and updated using UpdateCUDAProviderOptions.

Default value: 0

Performance Tuning

The I/O Binding feature should be utilized to avoid overhead resulting from copies on inputs and outputs. Ideally up and downloads for inputs can be hidden behind the inference. This can be achieved by doing asynchronous copies while running inference. This is demonstrated in this PR

Ort::RunOptions run_options;
run_options.AddConfigEntry("disable_synchronize_execution_providers", "1");
session->Run(run_options, io_binding);

By disabling the synchronization on the inference the user has to take care of synchronizing the compute stream after execution. This feature should only be used with device local memory or an ORT Value allocated in pinned memory, otherwise the issued download will be blocking and not behave as desired.

Convolution-heavy models

ORT leverages CuDNN for convolution operations and the first step in this process is to determine which “optimal” convolution algorithm to use while performing the convolution operation for the given input configuration (input shape, filter shape, etc.) in each Conv node. This sub-step involves querying CuDNN for a “workspace” memory size and have this allocated so that CuDNN can use this auxiliary memory while determining the “optimal” convolution algorithm to use.

The default value of cudnn_conv_use_max_workspace is 1 for versions 1.14 or later, and 0 for previous versions. When its value is 0, ORT clamps the workspace size to 32 MB which may lead to a sub-optimal convolution algorithm getting picked by CuDNN. To allow ORT to allocate the maximum possible workspace as determined by CuDNN, a provider option named cudnn_conv_use_max_workspace needs to get set (as shown below).

Keep in mind that using this flag may increase the peak memory usage by a factor (sometimes a few GBs) but this does help CuDNN pick the best convolution algorithm for the given input. We have found that this is an important flag to use while using an fp16 model as this allows CuDNN to pick tensor core algorithms for the convolution operations (if the hardware supports tensor core operations). This flag may or may not result in performance gains for other data types (float and double).

  • Python 
providers = [("CUDAExecutionProvider", {"cudnn_conv_use_max_workspace": '1'})]
  sess_options = ort.SessionOptions()
  sess = ort.InferenceSession("my_conv_heavy_fp16_model.onnx", sess_options=sess_options, providers=providers)
  • C/C++ 
OrtCUDAProviderOptionsV2* cuda_options = nullptr;
  CreateCUDAProviderOptions(&cuda_options);

  std::vector<const char*> keys{"cudnn_conv_use_max_workspace"};
  std::vector<const char*> values{"1"};

  UpdateCUDAProviderOptions(cuda_options, keys.data(), values.data(), 1);

  OrtSessionOptions* session_options = /* ... */;
  SessionOptionsAppendExecutionProvider_CUDA_V2(session_options, cuda_options);

  // Finally, don't forget to release the provider options
  ReleaseCUDAProviderOptions(cuda_options);
  • C# 
var cudaProviderOptions = new OrtCUDAProviderOptions(); // Dispose this finally

 var providerOptionsDict = new Dictionary<string, string>();
 providerOptionsDict["cudnn_conv_use_max_workspace"] = "1";

 cudaProviderOptions.UpdateOptions(providerOptionsDict);

 SessionOptions options = SessionOptions.MakeSessionOptionWithCudaProvider(cudaProviderOptions);  // Dispose this finally

Convolution Input Padding

ORT leverages CuDNN for convolution operations. While CuDNN only takes 4-D or 5-D tensor as input for convolution operations, dimension padding is needed if the input is 3-D tensor. Given an input tensor of shape [N, C, D], it can be padded to [N, C, D, 1] or [N, C, 1, D]. While both of these two padding ways produce same output, the performance may be a lot different because different convolution algorithms are selected, especially on some devices such as A100. By default the input is padded to [N, C, D, 1]. A provider option named cudnn_conv1d_pad_to_nc1d needs to get set (as shown below) if [N, C, 1, D] is preferred.

  • Python 
providers = [("CUDAExecutionProvider", {"cudnn_conv1d_pad_to_nc1d": '1'})]
  sess_options = ort.SessionOptions()
  sess = ort.InferenceSession("my_conv_model.onnx", sess_options=sess_options, providers=providers)
  • C/C++ 
OrtCUDAProviderOptionsV2* cuda_options = nullptr;
  CreateCUDAProviderOptions(&cuda_options);

  std::vector<const char*> keys{"cudnn_conv1d_pad_to_nc1d"};
  std::vector<const char*> values{"1"};

  UpdateCUDAProviderOptions(cuda_options, keys.data(), values.data(), 1);

  OrtSessionOptions* session_options = /* ... */;
  SessionOptionsAppendExecutionProvider_CUDA_V2(session_options, cuda_options);

  // Finally, don't forget to release the provider options
  ReleaseCUDAProviderOptions(cuda_options);
  • C# 
var cudaProviderOptions = new OrtCUDAProviderOptions(); // Dispose this finally

  var providerOptionsDict = new Dictionary<string, string>();
  providerOptionsDict["cudnn_conv1d_pad_to_nc1d"] = "1";

  cudaProviderOptions.UpdateOptions(providerOptionsDict);

  SessionOptions options = SessionOptions.MakeSessionOptionWithCudaProvider(cudaProviderOptions);  // Dispose this finally

Using CUDA Graphs (Preview)

While using the CUDA EP, ORT supports the usage of CUDA Graphs to remove CPU overhead associated with launching CUDA kernels sequentially. To enable the usage of CUDA Graphs, use the provider options as shown in the samples below. ORT supports multi-graph capture capability by passing the user specified gpu_graph_id to the run options. gpu_graph_id is optional when the session uses one cuda graph. If not set, the default value is 0. If the gpu_graph_id is set to -1, cuda graph capture/replay is disabled in that run.

Currently, there are some constraints with regards to using the CUDA Graphs feature:

  • Models with control-flow ops (i.e. If, Loop and Scan ops) are not supported.
  • Usage of CUDA Graphs is limited to models where-in all the model ops (graph nodes) can be partitioned to the CUDA EP.
  • The input/output types of models need to be tensors.
  • Shapes and addresses of inputs/outputs cannot change across inference calls for the same graph annotation id. Input tensors for replay shall be copied to the address of input tensors used in graph capture.
  • In multi-graph capture mode, the captured graphs will remain in the session’s lifetime and the captured graph deletion feature is not supported at the moment.
  • By design, CUDA Graphs is designed to read from/write to the same CUDA virtual memory addresses during the graph replaying step as it does during the graph capturing step. Due to this requirement, usage of this feature requires using IOBinding so as to bind memory which will be used as input(s)/output(s) for the CUDA Graph machinery to read from/write to (please see samples below).
  • While updating the input(s) for subsequent inference calls, the fresh input(s) need to be copied over to the corresponding CUDA memory location(s) of the bound OrtValue input(s) (please see samples below to see how this can be achieved). This is due to the fact that the “graph replay” will require reading inputs from the same CUDA virtual memory addresses.
  • Multi-threaded usage is currently not supported, i.e. Run() MAY NOT be invoked on the same InferenceSession object from multiple threads while using CUDA Graphs.

NOTE: The very first Run() performs a variety of tasks under the hood like making CUDA memory allocations, capturing the CUDA graph for the model, and then performing a graph replay to ensure that the graph runs. Due to this, the latency associated with the first Run() is bound to be high. Subsequent Run()s only perform graph replays of the graph captured and cached in the first Run().

  • Python
providers = [("CUDAExecutionProvider", {"enable_cuda_graph": '1'})]
  sess_options = ort.SessionOptions()
  sess = ort.InferenceSession("my_model.onnx", sess_options=sess_options, providers=providers)

  providers = [("CUDAExecutionProvider", {'enable_cuda_graph': True})]
  x = np.array([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]], dtype=np.float32)
  y = np.array([[0.0], [0.0], [0.0]], dtype=np.float32)
  x_ortvalue = onnxrt.OrtValue.ortvalue_from_numpy(x, 'cuda', 0)
  y_ortvalue = onnxrt.OrtValue.ortvalue_from_numpy(y, 'cuda', 0)

  session = onnxrt.InferenceSession("matmul_2.onnx", providers=providers)
  io_binding = session.io_binding()

  # Pass gpu_graph_id to RunOptions through RunConfigs
  ro = onnxrt.RunOptions()
  # gpu_graph_id is optional if the session uses only one cuda graph
  ro.add_run_config_entry("gpu_graph_id", "1")

  # Bind the input and output
  io_binding.bind_ortvalue_input('X', x_ortvalue)
  io_binding.bind_ortvalue_output('Y', y_ortvalue)

  # One regular run for the necessary memory allocation and cuda graph capturing
  session.run_with_iobinding(io_binding, ro)
  expected_y = np.array([[5.0], [11.0], [17.0]], dtype=np.float32)
  np.testing.assert_allclose(expected_y, y_ortvalue.numpy(), rtol=1e-05, atol=1e-05)

  # After capturing, CUDA graph replay happens from this Run onwards
  session.run_with_iobinding(io_binding, ro)
  np.testing.assert_allclose(expected_y, y_ortvalue.numpy(), rtol=1e-05, atol=1e-05)

  # Update input and then replay CUDA graph with the updated input
  x_ortvalue.update_inplace(np.array([[10.0, 20.0], [30.0, 40.0], [50.0, 60.0]], dtype=np.float32))
  session.run_with_iobinding(io_binding, ro)
  • C/C++ 
const auto& api = Ort::GetApi();

  struct CudaMemoryDeleter {
  explicit CudaMemoryDeleter(const Ort::Allocator* alloc) {
      alloc_ = alloc;
  }

  void operator()(void* ptr) const {
      alloc_->Free(ptr);
  }

  const Ort::Allocator* alloc_;
  };

  // Enable cuda graph in cuda provider option.
  OrtCUDAProviderOptionsV2* cuda_options = nullptr;
  api.CreateCUDAProviderOptions(&cuda_options);
  std::unique_ptr<OrtCUDAProviderOptionsV2, decltype(api.ReleaseCUDAProviderOptions)> rel_cuda_options(cuda_options, api.ReleaseCUDAProviderOptions);
  std::vector<const char*> keys{"enable_cuda_graph"};
  std::vector<const char*> values{"1"};
  api.UpdateCUDAProviderOptions(rel_cuda_options.get(), keys.data(), values.data(), 1);

  Ort::SessionOptions session_options;
  api.SessionOptionsAppendExecutionProvider_CUDA_V2(static_cast<OrtSessionOptions*>(session_options), rel_cuda_options.get();

  // Pass gpu_graph_id to RunOptions through RunConfigs
  Ort::RunOptions run_option;
  // gpu_graph_id is optional if the session uses only one cuda graph
  run_option.AddConfigEntry("gpu_graph_id", "1");

  // Create IO bound inputs and outputs.
  Ort::Session session(*ort_env, ORT_TSTR("matmul_2.onnx"), session_options);
  Ort::MemoryInfo info_cuda("Cuda", OrtAllocatorType::OrtArenaAllocator, 0, OrtMemTypeDefault);
  Ort::Allocator cuda_allocator(session, info_cuda);

  const std::array<int64_t, 2> x_shape = {3, 2};
  std::array<float, 3 * 2> x_values = {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f};
  auto input_data = std::unique_ptr<void, CudaMemoryDeleter>(cuda_allocator.Alloc(x_values.size() * sizeof(float)),
                                                          CudaMemoryDeleter(&cuda_allocator));
  cudaMemcpy(input_data.get(), x_values.data(), sizeof(float) * x_values.size(), cudaMemcpyHostToDevice);

  // Create an OrtValue tensor backed by data on CUDA memory
  Ort::Value bound_x = Ort::Value::CreateTensor(info_cuda, reinterpret_cast<float*>(input_data.get()), x_values.size(),
                                              x_shape.data(), x_shape.size());

  const std::array<int64_t, 2> expected_y_shape = {3, 2};
  std::array<float, 3 * 2> expected_y = {1.0f, 4.0f, 9.0f, 16.0f, 25.0f, 36.0f};
  auto output_data = std::unique_ptr<void, CudaMemoryDeleter>(cuda_allocator.Alloc(expected_y.size() * sizeof(float)),
                                                              CudaMemoryDeleter(&cuda_allocator));

  // Create an OrtValue tensor backed by data on CUDA memory
  Ort::Value bound_y = Ort::Value::CreateTensor(info_cuda, reinterpret_cast<float*>(output_data.get()),
                                              expected_y.size(), expected_y_shape.data(), expected_y_shape.size());

  Ort::IoBinding binding(session);
  binding.BindInput("X", bound_x);
  binding.BindOutput("Y", bound_y);

  // One regular run for necessary memory allocation and graph capturing
  session.Run(run_option, binding);

  // After capturing, CUDA graph replay happens from this Run onwards
  session.Run(run_option, binding);

  // Update input and then replay CUDA graph with the updated input
  x_values = {10.0f, 20.0f, 30.0f, 40.0f, 50.0f, 60.0f};
  cudaMemcpy(input_data.get(), x_values.data(), sizeof(float) * x_values.size(), cudaMemcpyHostToDevice);
  session.Run(run_option, binding);
  • C# (future)

Samples

Python

import onnxruntime as ort

model_path = '<path to model>'

providers = [
    ('CUDAExecutionProvider', {
        'device_id': 0,
        'arena_extend_strategy': 'kNextPowerOfTwo',
        'gpu_mem_limit': 2 * 1024 * 1024 * 1024,
        'cudnn_conv_algo_search': 'EXHAUSTIVE',
        'do_copy_in_default_stream': True,
    }),
    'CPUExecutionProvider',
]

session = ort.InferenceSession(model_path, providers=providers)

C/C++

USING LEGACY PROVIDER OPTIONS STRUCT
OrtSessionOptions* session_options = /* ... */;

OrtCUDAProviderOptions options;
options.device_id = 0;
options.arena_extend_strategy = 0;
options.gpu_mem_limit = 2 * 1024 * 1024 * 1024;
options.cudnn_conv_algo_search = OrtCudnnConvAlgoSearchExhaustive;
options.do_copy_in_default_stream = 1;

SessionOptionsAppendExecutionProvider_CUDA(session_options, &options);
USING V2 PROVIDER OPTIONS STRUCT
OrtCUDAProviderOptionsV2* cuda_options = nullptr;
CreateCUDAProviderOptions(&cuda_options);

std::vector<const char*> keys{"device_id", "gpu_mem_limit", "arena_extend_strategy", "cudnn_conv_algo_search", "do_copy_in_default_stream", "cudnn_conv_use_max_workspace", "cudnn_conv1d_pad_to_nc1d"};
std::vector<const char*> values{"0", "2147483648", "kSameAsRequested", "DEFAULT", "1", "1", "1"};

UpdateCUDAProviderOptions(cuda_options, keys.data(), values.data(), keys.size());

cudaStream_t cuda_stream;
cudaStreamCreate(&cuda_stream);
// this implicitly sets "has_user_compute_stream"
UpdateCUDAProviderOptionsWithValue(cuda_options, "user_compute_stream", cuda_stream);
OrtSessionOptions* session_options = /* ... */;
SessionOptionsAppendExecutionProvider_CUDA_V2(session_options, cuda_options);

// Finally, don't forget to release the provider options
ReleaseCUDAProviderOptions(cuda_options);

C#

var cudaProviderOptions = new OrtCUDAProviderOptions(); // Dispose this finally

var providerOptionsDict = new Dictionary<string, string>();
providerOptionsDict["device_id"] = "0";
providerOptionsDict["gpu_mem_limit"] = "2147483648";
providerOptionsDict["arena_extend_strategy"] = "kSameAsRequested";
providerOptionsDict["cudnn_conv_algo_search"] = "DEFAULT";
providerOptionsDict["do_copy_in_default_stream"] = "1";
providerOptionsDict["cudnn_conv_use_max_workspace"] = "1";
providerOptionsDict["cudnn_conv1d_pad_to_nc1d"] = "1";

cudaProviderOptions.UpdateOptions(providerOptionsDict);

SessionOptions options = SessionOptions.MakeSessionOptionWithCudaProvider(cudaProviderOptions);  // Dispose this finally

Also see the tutorial here on how to configure CUDA for C# on Windows.

Java

OrtCUDAProviderOptions cudaProviderOptions = new OrtCUDAProviderOptions(/*device id*/0); // Must be closed after the session closes

cudaProviderOptions.add("gpu_mem_limit","2147483648");
cudaProviderOptions.add("arena_extend_strategy","kSameAsRequested");
cudaProviderOptions.add("cudnn_conv_algo_search","DEFAULT");
cudaProviderOptions.add("do_copy_in_default_stream","1");
cudaProviderOptions.add("cudnn_conv_use_max_workspace","1");
cudaProviderOptions.add("cudnn_conv1d_pad_to_nc1d","1");

OrtSession.SessionOptions options = new OrtSession.SessionOptions(); // Must be closed after the session closes
options.addCUDA(cudaProviderOptions);

For documentation questions, please file an issue.