onnx-tensorrt icon indicating copy to clipboard operation
onnx-tensorrt copied to clipboard

tensorrt use onnx_parser to parse onnx file error

Open joan126 opened this issue 3 years ago • 1 comments

Description

error occured when use tensorrt to build engine with onnx model file using onnx_parser .

onnx model from tensorflow-gpu 1.14.0

Environment

TensorRT Version: 8.2.3 ONNX-TensorRT Version / Branch: latest GPU Type: TITAN X Nvidia Driver Version: 440.100 CUDA Version: 10.2. CUDNN Version: 7.6.5 Operating System + Version: ubuntu18.04 Python Version (if applicable): 3.7 TensorFlow + TF2ONNX Version (if applicable): 1.14.0 + 1.9.0 PyTorch Version (if applicable): None Baremetal or Container (if container which image + tag):

Relevant Files

[02/17/2022-08:37:07] [TRT] [V] Registering layer: Tacotron_model/inference/decoder/while/LogicalAnd_1 for ONNX node: Tacotron_model/inference/decoder/while/LogicalAnd_1 [02/17/2022-08:37:07] [TRT] [V] Registering tensor: sub_graph_ending_node_Identity__303:0 for ONNX tensor: sub_graph_ending_node_Identity__303:0 [02/17/2022-08:37:07] [TRT] [V] Tacotron_model/inference/decoder/while/LogicalAnd_1 [And] outputs: [sub_graph_ending_node_Identity__303:0 -> ()[BOOL]], [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__304:0, found matching tensor: sub_graph_ending_node_Identity__304:0, with shape: (1, -1) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__305:0, found matching tensor: sub_graph_ending_node_Identity__305:0, with shape: (1, 1024) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__306:0, found matching tensor: sub_graph_ending_node_Identity__306:0, with shape: (1) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__307:0, found matching tensor: sub_graph_ending_node_Identity__307:0, with shape: (1, 1024) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__308:0, found matching tensor: sub_graph_ending_node_Identity__308:0, with shape: (1, 1024) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__309:0, found matching tensor: sub_graph_ending_node_Identity__309:0, with shape: () [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__310:0, found matching tensor: sub_graph_ending_node_Identity__310:0, with shape: (1, 544) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__311:0, found matching tensor: sub_graph_ending_node_Identity__311:0, with shape: (1, 1024) [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__312:0, found matching tensor: sub_graph_ending_node_Identity__312:0, with shape: () [02/17/2022-08:37:07] [TRT] [V] For state variable output: sub_graph_ending_node_Identity__313:0, found matching tensor: sub_graph_ending_node_Identity__313:0, with shape: (1, 80) [02/17/2022-08:37:07] [TRT] [V] For scan output: sub_graph_ending_node_Identity__314:0, found matching tensor: sub_graph_ending_node_Identity__314:0, with shape: (1, 2) [02/17/2022-08:37:07] [TRT] [E] [graphShapeAnalyzer.cpp::processCheck::581] Error Code 4: Internal Error ((Unnamed Layer* 331) [LoopOutput]_output: tensor volume exceeds (2^31)-1, dimensions are [2147483647,1,2]) ERROR: Failed to parse the ONNX file. In node 30 (parseGraph): INVALID_NODE: Invalid Node - generic_loop_Loop__333 [graphShapeAnalyzer.cpp::processCheck::581] Error Code 4: Internal Error ((Unnamed Layer* 331) [LoopOutput]_output: tensor volume exceeds (2^31)-1, dimensions are [2147483647,1,2])

joan126 avatar Feb 17 '22 08:02 joan126

Are you able the share the model?

kevinch-nv avatar Feb 22 '22 06:02 kevinch-nv