Caffe2 (tag v0.4.1) c++ api, can not run with GPU support.

Autonomous Machines Jetson & Embedded Systems Jetson TX1
1

Hi, I use Jetpack 3.2.1, build caffe2 from source (tag v0.4.1), cuda9, cudnn7 on TX1.
I compiled and installed caffe2 successfully, and I run python caffe2 works well with GPU. Then I try to run caffe2 in C++ with GPU support, but c++ app runs ok, but no GPU usage. This code works well on my laptop with GPU support, I don’t know why, here is my code(https://github.com/zacario-li/caffe2_Cpp_example), hope anyone could help me. Many thanks.

2

Hi,

Looks like this issue is duplicate to this comment:
[url]https://devtalk.nvidia.com/default/topic/1030538/jetson-tx2/caffe2-on-jetson-tx2-gpu-support-/post/5274178/#5274178[/url]

JetPack3.1 includes CUDA 8.0.
If you want to have a CUDA 9.0 environment, please upgrade your JetPack to version 3.2.1/3.3.
[url]https://developer.nvidia.com/embedded/jetpack-archive[/url]

Thanks.

3

Sorry for my mistake, my jetpack version is v3.2.1.

4

Hi,

Thanks for the feedback.

We are try to reproduce this issue internally.
Will update information with you later.

5

Hi,

We try to compile caffe2 to reproduce this issue but meet some error when cmaking.
Ex. Unknown CMake command “EXCLUDE” and “caffe2_interface_library”

Suppose you built caffe2 from source, could you share your steps with us?
So that we can check the GPU utilization issue directly.

Thanks.

6

hi,
what’s your build steps? I didn’t meet this issue. I just followed the offical website build guide. The build approximately took me 50 mins to finish it.

7

I use this offical repository(GitHub - pytorch/pytorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration),tag v0.4.1

8

Hi,

Not sure if you have compiled the correct architecture for TX1.
Could you echo the NVCC_FLAGS_EXTRA to check if sm_53 is well-included?
[url]pytorch/cuda.cmake at master · pytorch/pytorch · GitHub

Thanks.