05-27-2019 04:28 AM
I am working under the user guide of the DNNDK.
I can see that DNNC is installed successfully.
$ dnnc --version
dnnc version v2.04
DPU Target : v1.4.0
Build Label: Apr 24 2019 19:12:00
Copyright @2019 Xilinx Inc. All Rights Reserved.
However, there is a problem when I tried to install the Deep Compression Tool (DECENT). The command and error can be seen below.
$ decent --version
decent: error while loading shared libraries: libcudart.so.9.0: cannot open shared object file: No such file or directory
Can you please help me check this issue? Thank you!
05-27-2019 07:18 PM
05-27-2019 06:34 PM
The message "error while loading shared libraries: libcudart.so.9.0" indicates that a CUDA library is missing. Please see UG1327 (v1.4) p. 8:
You may try using your favorite search engine to look for instructions on how to install CUDA and cuDNN using the keywords "install cudnn".
05-27-2019 07:10 PM
Thanks for your detailed reply!
I have thought about this possibility. However, it seems that I have already installed the required packages. Below is the command and information when I was trying to install DECENT and DNNC.
$ sudo ./install.sh ZCU102
Inspect system environment ...
Begin to install Xilinx DNNDK tools on host ...
Complete dnnc installation successfully.
Complete CPU version of decent for caffe installation successfully.
Complete GPU version of decent installation successfully.
I have used the following commands to check the CUDA and cuDNN version. Any idea on what is the problem?
$ cat /usr/local/cuda/version.txt
CUDA Version 9.0.176
$ cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
#define CUDNN_MAJOR 7
#define CUDNN_MINOR 0
#define CUDNN_PATCHLEVEL 5
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)
05-27-2019 07:18 PM
05-23-2021 04:25 AM
I also have a similar question. I'm using Ubuntu 18.04 and Cuda 10.0 Cudnn 7.4.1, I can run xilinx_dnndk_v3.1/host_x86/install.sh and install caffe GPU version successfully. However, if I run 'decent --version' in other folders except '/usr/local/cuda/lib64/', there is 'decent: error while loading shared libraries: libcudart.so.10.0: cannot open shared object file: No such file or directory', and I check 'libcudart.so.10.0' is in '/usr/local/lib64/', if I run 'decent --version' in 'usr/local/cuda/lib64', there is no fault, I want to know why.
06-14-2021 11:51 AM
Even I have a similar issue, I am unable to install DNNDK Host Tools package as illustrated on ug1327 page 9. I get some kind of error message while I install it. I'm thinking that I should initially install CUDA 9.0 and cuDNN 7.0.5. after this maybe I should focus on the main installation. Let me know if you think will work. Basically, this is to develop deep learning applications for Enterprise Application Development Services and I hope to cover the phases of neural network model compression, programming, compilation, and runtime enablement. cheers