OutOfMemory Error in computeCosts: 0

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Hello guys!

I ran into following error when building the engine from UFF file.
Also I tried to modify builder.max_batch_size to xK,xM,xG…
But nothing has changed

Any advise please

thanks

Platform info:
Jetson tx2
CUDA Version 10.0.166
Python 3.6.7
TensorRT version 5.0.6.3

Error:

[TensorRT] ERROR: Internal error: could not find any implementation for node (Unnamed Layer* 484) [Deconvolution], try increasing the workspace size with IBuilder::setMaxWorkspaceSize()
[TensorRT] ERROR: …/builder/tacticOptimizer.cpp (1230) - OutOfMemory Error in computeCosts: 0
Traceback (most recent call last):
File “trt.py”, line 191, in
make_trt_engine()
File “trt.py”, line 119, in make_trt_engine
with builder.build_cuda_engine(network) as engine:
AttributeError: enter

Code:

def make_trt_engine():
    with trt.Builder(TRT_LOGGER) as builder, builder.create_network() as network, trt.UffParser() as parser:
        parser.register_input("Placeholder", (1,256,256,3))
        parser.register_output("sigmoid")
        parser.parse("model.uff", network)
    
        builder.max_workspace_size = 1<<30
        builder.max_batch_size = 1
        with builder.build_cuda_engine(network) as engine:
            with open("model.engine", "wb") as f:
                f.write(engine.serialize())

return
2

My fault
I executed the code twice.

with trt.Builder(TRT_LOGGER) as builder