I am using Ubuntu 18.04.1 LTS I installed cuda-9.0, Pytorch 0.4, libcupti and cuDNN 7.0 for deep learning purposes.
I got error:
ImportError: libcublas.so.9.0: cannot open shared object file: No such file or directory
GPU: GeForce GTX 1050/PCIe/SSE2
| NVIDIA-SMI 415.23 Driver Version: 415.23 CUDA Version: 10.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1050 Off | 00000000:01:00.0 Off | N/A |
| N/A 49C P5 N/A / N/A | 474MiB / 4042MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 1104 G /usr/lib/xorg/Xorg 28MiB |
| 0 1198 G /usr/bin/gnome-shell 48MiB |
| 0 1384 G /usr/lib/xorg/Xorg 254MiB |
| 0 1571 G /usr/bin/gnome-shell 137MiB |
| 0 2148 G /usr/lib/firefox/firefox 1MiB |
| 0 3136 G /usr/lib/firefox/firefox 1MiB |
| 0 7041 G gnome-control-center 1MiB |
+-----------------------------------------------------------------------------+
test gpu install
python -c 'import torch; print(torch.rand(2,3).cuda())'
tensor([[0.9905, 0.9057, 0.2803],
[0.3525, 0.1388, 0.0692]], device='cuda:0')
I checked Python environment is already configured:
Python 3.6.6 :: Anaconda, Inc.
pip 18.1 from /home/wojtek/anaconda3/lib/python3.6/site-packages/pip (python 3.6)
virtualenv --version
16.1.0
I installed tensorflow-gpu using conda.
test gpu install:
python -c "import tensorflow as tf; tf.enable_eager_execution(); print(tf.reduce_sum(tf.random_normal([1000, 1000])))"
2018-12-17 16:13:24.226023: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2018-12-17 16:13:24.307372: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:964] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-12-17 16:13:24.307737: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1432] Found device 0 with properties:
name: GeForce GTX 1050 major: 6 minor: 1 memoryClockRate(GHz): 1.493
pciBusID: 0000:01:00.0
totalMemory: 3.95GiB freeMemory: 3.61GiB
2018-12-17 16:13:24.307750: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1511] Adding visible gpu devices: 0
2018-12-17 16:13:24.488105: I tensorflow/core/common_runtime/gpu/gpu_device.cc:982] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-12-17 16:13:24.488132: I tensorflow/core/common_runtime/gpu/gpu_device.cc:988] 0
2018-12-17 16:13:24.488137: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1001] 0: N
2018-12-17 16:13:24.488337: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 3329 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1050, pci bus id: 0000:01:00.0, compute capability: 6.1)
tf.Tensor(471.4757, shape=(), dtype=float32)
echo $LD_LIBRARY_PATH
/usr/local/cuda/lib64:
(base) wojtek@wojtek-GF63-8RC:~$ ls /usr/local/cuda-9.0/lib64
libaccinj64.so libcusparse.so libnppist.so.9.0
libaccinj64.so.9.0 libcusparse.so.9.0 libnppist.so.9.0.176
libaccinj64.so.9.0.176 libcusparse.so.9.0.176 libnppist_static.a
libcublas_device.a libcusparse_static.a libnppisu.so
libcublas.so libnppc.so libnppisu.so.9.0
libcublas.so.9.0 libnppc.so.9.0 libnppisu.so.9.0.176
libcublas.so.9.0.176 libnppc.so.9.0.176 libnppisu_static.a
libcublas_static.a libnppc_static.a libnppitc.so
libcudadevrt.a libnppial.so libnppitc.so.9.0
libcudart.so libnppial.so.9.0 libnppitc.so.9.0.176
libcudart.so.9.0 libnppial.so.9.0.176 libnppitc_static.a
libcudart.so.9.0.176 libnppial_static.a libnpps.so
libcudart_static.a libnppicc.so libnpps.so.9.0
libcudnn.so libnppicc.so.9.0 libnpps.so.9.0.176
libcudnn.so.7 libnppicc.so.9.0.176 libnpps_static.a
libcudnn.so.7.0.5 libnppicc_static.a libnvblas.so
libcudnn_static.a libnppicom.so libnvblas.so.9.0
libcufft.so libnppicom.so.9.0 libnvblas.so.9.0.176
libcufft.so.9.0 libnppicom.so.9.0.176 libnvgraph.so
libcufft.so.9.0.176 libnppicom_static.a libnvgraph.so.9.0
libcufft_static.a libnppidei.so libnvgraph.so.9.0.176
libcufftw.so libnppidei.so.9.0 libnvgraph_static.a
libcufftw.so.9.0 libnppidei.so.9.0.176 libnvrtc-builtins.so
libcufftw.so.9.0.176 libnppidei_static.a libnvrtc-builtins.so.9.0
libcufftw_static.a libnppif.so libnvrtc-builtins.so.9.0.176
libcuinj64.so libnppif.so.9.0 libnvrtc.so
libcuinj64.so.9.0 libnppif.so.9.0.176 libnvrtc.so.9.0
libcuinj64.so.9.0.176 libnppif_static.a libnvrtc.so.9.0.176
libculibos.a libnppig.so libnvToolsExt.so
libcurand.so libnppig.so.9.0 libnvToolsExt.so.1
libcurand.so.9.0 libnppig.so.9.0.176 libnvToolsExt.so.1.0.0
libcurand.so.9.0.176 libnppig_static.a libOpenCL.so
libcurand_static.a libnppim.so libOpenCL.so.1
libcusolver.so libnppim.so.9.0 libOpenCL.so.1.0
libcusolver.so.9.0 libnppim.so.9.0.176 libOpenCL.so.1.0.0
libcusolver.so.9.0.176 libnppim_static.a stubs
libcusolver_static.a libnppist.so
Thanks in advance.