[TensorRT] ERROR: Internal error: could not find any implementation for node (Unnamed Layer* 202) [Deconvolution], try increasing the workspace size with IBuilder::setMaxWorkspaceSize()

[ShreyasSkandanS]

[TensorRT] ERROR: Internal error: could not find any implementation for node (Unnamed Layer* 202) [Deconvolution], try increasing the workspace size with IBuilder::setMaxWorkspaceSize()
[TensorRT] ERROR: ../builder/tacticOptimizer.cpp (1230) - OutOfMemory Error in computeCosts: 0

I have a network (based on ERFNet) and some trained weights for this model. I'm trying to convert it from PyTorch to TensorRT and I do the following:

import torch
from torch2trt import torch2trt
import pdb

from NETWORK_DESIGN import Net
NUM_CLASSES = X
model = Net(NUM_CLASSES)
model_url = torch.load("/path/to/model/model_best.pth")
model.load_state_dict(model_url, strict=False)
model.cuda().half().eval()
x = torch.ones((1, 3, WIDTH, HEIGHT)).cuda().half()
model_trt = torch2trt(model, [x], fp16_mode=True)
pdb.set_trace()

But I get the above error. Is there an unsupported layer in here somewhere? I don't believe there to be any fancy layers in this architecture. I'm happy to share both the network design and weights files. I tested this both on my Laptop (MX150) and a Jetson Xavier (Jetpack 4.2).

The network was trained using PyTorch 1.1.

[jaybdub]

Hi ShreyasSkandanS,

Thanks for reaching out. Could you share the version of TensorRT you are using?

Also, can provide the PyTorch model you're using? Sometimes the solution can be a bit nuanced.

The trained weights likely aren't necessary.

Best, John

[ShreyasSkandanS]

I would prefer to share the model + weights offline, if that is possible?

[jaybdub]

Sure, jwelsh@nvidia.com

[ShreyasSkandanS]

Silly me, I believe I just reached out to you.

[ShreyasSkandanS]

Solved.

[allenling]

i got same problem, any help?

[pwr617]

i got same problem, any help?

[jaybdub]

Hi all,

I believe the solution was to increase max workspace size. You can do this by setting max_workspace_size parameter. For example

‘’model_trt = torch2trt(model, [data], max_workspace_size=1<<25)’’

Should work.

Best, John

[BurkeyLai]

Hi all,

I believe the solution was to increase max workspace size. You can do this by setting max_workspace_size parameter. For example

‘’model_trt = torch2trt(model, [data], max_workspace_size=1<<25)’’

Should work.

Best, John

This answer really help me a lot. Thanks!