easy-tensorrt-sys 0.2.1

Rust binding to NVIDIA TensorRT, forked from tensorrt-rs-sys
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163

#[cxx::bridge]
pub(crate) mod ffi {
    #[namespace = "trt_rs::logger"]
    unsafe extern "C++" {
        include!("easy-tensorrt-sys/cxx/include/logger.h");

        type Logger;

        fn create_logger() -> UniquePtr<Logger>;

        fn log(self: Pin<&mut Logger>, severity: i32, msg: &str);

        fn set_level(self: Pin<&mut Logger>, severity: i32);
    }

    #[namespace = "trt_rs::runtime"]
    unsafe extern "C++" {
        include!("easy-tensorrt-sys/cxx/include/runtime.h");

        type Runtime;

        type CudaEngine;

        type ExecutionContext;

        // Runtime
        fn create_runtime(logger: Pin<&mut Logger>) -> UniquePtr<Runtime>;

        fn deserialize(self: Pin<&mut Runtime>, data: &[u8]) -> UniquePtr<CudaEngine>;

        fn set_max_threads(self: Pin<&mut Runtime>, max_threads: i32) -> bool;

        fn get_max_threads(self: &Runtime) -> i32;

        fn set_engine_host_code_allowed(self: Pin<&mut Runtime>, allowed: bool);

        fn get_engine_host_code_allowed(self: &Runtime) -> bool;

        // CudaEngine
        fn get_tensor_shape(self: &CudaEngine, name: &str) -> Vec<i32>;

        fn get_tensor_dtype(self: &CudaEngine, name: &str) -> i32;

        fn get_num_layers(self: &CudaEngine) -> i32;

        fn create_execution_context(self: Pin<&mut CudaEngine>) -> UniquePtr<ExecutionContext>;

        fn is_shape_inference_io(self: &CudaEngine, name: &str) -> bool;

        fn get_tensor_io_mode(self: &CudaEngine, name: &str) -> i32;

        fn create_execution_context_without_device_memory(
            self: Pin<&mut CudaEngine>,
        ) -> UniquePtr<ExecutionContext>;

        fn get_device_memory_size(self: &CudaEngine) -> usize;

        fn is_refittable(self: &CudaEngine) -> bool;

        fn get_tensor_bytes_per_component(self: &CudaEngine, name: &str) -> i32;

        fn get_tensor_components_per_element(self: &CudaEngine, name: &str) -> i32;

        fn get_tensor_format(self: &CudaEngine, name: &str) -> i32;

        fn get_tensor_vectorized_dim(self: &CudaEngine, name: &str) -> i32;

        fn get_name(self: &CudaEngine) -> &str;

        fn get_num_optimization_profiles(self: &CudaEngine) -> i32;

        fn get_engine_capability(self: &CudaEngine) -> i32;

        fn has_implicit_batch_dimension(self: &CudaEngine) -> bool;

        fn get_num_io_tensors(self: &CudaEngine) -> i32;

        fn get_io_tensor_name(self: &CudaEngine, index: i32) -> &str;

        fn get_hardware_compatibility_level(self: &CudaEngine) -> i32;

        fn get_num_aux_streams(self: &CudaEngine) -> i32;

        // ExecutionContext
        fn set_debug_sync(self: Pin<&mut ExecutionContext>, sync: bool);

        fn get_debug_sync(self: &ExecutionContext) -> bool;

        fn set_name(self: Pin<&mut ExecutionContext>, name: &str);

        fn get_name(self: &ExecutionContext) -> &str;

        fn set_device_memory(self: Pin<&mut ExecutionContext>, memory: usize);

        fn get_tensor_strides(self: &ExecutionContext, name: &str) -> Vec<i32>;

        fn get_optimization_profile(self: &ExecutionContext) -> i32;

        fn set_input_shape(self: Pin<&mut ExecutionContext>, name: &str, shape: &[i32]) -> bool;

        fn get_tensor_shape(self: &ExecutionContext, name: &str) -> Vec<i32>;

        fn all_input_dimensions_specified(self: &ExecutionContext) -> bool;

        fn all_input_shapes_specified(self: &ExecutionContext) -> bool;

        fn set_optimization_profile_async(
            self: Pin<&mut ExecutionContext>,
            profile_index: i32,
            stream: usize,
        ) -> bool;

        fn set_enqueue_emits_profile(self: Pin<&mut ExecutionContext>, emits: bool);

        fn get_enqueue_emits_profile(self: &ExecutionContext) -> bool;

        fn report_to_profiler(self: Pin<&mut ExecutionContext>);

        fn set_tensor_address(self: Pin<&mut ExecutionContext>, name: &str, address: usize)
            -> bool;

        fn get_tensor_address(self: &ExecutionContext, name: &str) -> usize;

        fn set_input_tensor_address(
            self: Pin<&mut ExecutionContext>,
            name: &str,
            address: usize,
        ) -> bool;

        fn get_output_tensor_address(self: &ExecutionContext, name: &str) -> usize;

        fn set_input_consumed_event(self: Pin<&mut ExecutionContext>, event: usize) -> bool;

        fn get_input_consumed_event(self: &ExecutionContext) -> usize;

        fn get_max_output_size(self: &ExecutionContext, name: &str) -> usize;

        fn enqueue_v3(self: Pin<&mut ExecutionContext>, stream: usize) -> bool;

        fn set_persistent_cache_limit(self: Pin<&mut ExecutionContext>, limit: usize);

        fn get_persistent_cache_limit(self: &ExecutionContext) -> usize;

        fn set_nvtx_verbosity(self: Pin<&mut ExecutionContext>, verbosity: i32);

        fn set_aux_streams(self: Pin<&mut ExecutionContext>, streams: &[usize]);
    }

    #[namespace = "trt_rs::plugin"]
    unsafe extern "C++" {
        include!("easy-tensorrt-sys/cxx/include/plugin.h");

        fn init_libnv_infer_plugins() -> bool;

        fn load_library(plugin_path: &str) -> usize;

        fn unload_library(handle: usize);
    }
}

pub mod logger;
pub mod plugin;
pub mod runtime;