Driver error when addition or multiplication more than 2 times

[Lamply]

Test code:

import torch
import torch.nn as nn
from torch2trt import torch2trt

class TestModel(nn.Module):
    def __init__(self):
        super(TestModel,self).__init__()
    
    def forward(self, x):
        return x+x+x

# Only occur when using cuda
model = TestModel().cuda().eval()
x = torch.rand((1, 3, 224, 224)).cuda()
model_trt = torch2trt(model, [x])

Log:

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-5-ca2b186b88d6> in <module>
      8 model = TestModel().cuda().eval()
      9 x = torch.rand((1, 3, 224, 224)).cuda()
---> 10 model_trt = torch2trt(model, [x])

/usr/local/lib/python3.6/dist-packages/torch2trt-0.1.0-py3.6-linux-x86_64.egg/torch2trt/torch2trt.py in torch2trt(module, inputs, input_names, output_names, log_level, max_batch_size, fp16_mode, max_workspace_size, strict_type_constraints, keep_network, int8_mode, int8_calib_dataset, int8_calib_algorithm, int8_calib_batch_size, use_onnx)
    553         )
    554 
--> 555     engine = builder.build_cuda_engine(network)
    556 
    557     module_trt = TRTModule(engine, input_names, output_names)

RuntimeError: Driver error: 

Environment:

• Ubuntu18.04
• Python3.6.9
• CUDA 10.2
• cuDNN 8.0
• TensorRT 7.2.1
• torch2trt 0.1.0

[neoyang0620]

I met the same bug. Did you find out the solution?

[abaumhof]

I hit the same error when going through the example notebook at trt_pose (https://github.com/NVIDIA-AI-IOT/trt_pose/blob/master/tasks/human_pose/live_demo.ipynb)

I also tried your program and got the same issue. I'm on Ubuntu 18.04, python 3.6.9, CUDA 11.1 and TensorRT 7.2.3

Did anyone find a solution?

[ygy12345678]

I met the same bug

[Lamply]

@neoyang0620 @abaumhof @ygy12345678 FYI, Torch-TensorRT(https://github.com/NVIDIA/Torch-TensorRT) might be a better way to convert PyTorch models to TensorRT, or simply insert a dummy layer between them to avoid this issue.

[cxsweet]

想问一下后面这个问题是怎么解决的啊（卡好久了