GPU->CPU Memcpy failed (pip version of tensorflow)

NifTK / NiftyNet

[unmaintained] An open-source convolutional neural networks platform for research in medical image analysis and image-guided therapy

http://niftynet.io

Apache License 2.0

1.36k stars 404 forks source link

GPU->CPU Memcpy failed (pip version of tensorflow) #445

Closed IsaacLord closed 4 years ago

IsaacLord commented 4 years ago

I'm trying to use 3D network by using GPU (2080 Ti), however i got following error.

GPU->CPU Memcpy failed

2019-10-01 11:06:06.793519: I tensorflow/stream_executor/stream.cc:5011] [stream=00000256E1A01E70,impl=00000256E5F61A70] did not memcpy device-to-host; source: 0000000B11880D00 2019-10-01 11:06:06.791243: F tensorflow/core/common_runtime/gpu/gpu_util.cc:292] GPU->CPU Memcpy failed

even with spatial window (8,8,8) by batch size=1.

is there any idea how to fix this?

IsaacLord commented 4 years ago

Just change the Tensorflow Version!

wyli commented 4 years ago

thanks @DeepMedi probably https://github.com/NifTK/NiftyNet/blob/dev/requirements-gpu.txt#L6 needs to be revised due to some recent changes in tensorflow's pip package

talmazov commented 4 years ago

Just change the Tensorflow Version!

What tensorflow version and numpy did you use? i have niftynet 0.6, CUDA 10.0, tensorflow-gpu 1.13.2 and numpy 1.16 using geforce RTX 2060 6GB vram with nvidia driver 440.33.01

Still doesn’t work...