cudnn problems

hi,everyone,I am using Manjaro.

i got a cuda9.0 on my device.so I download cuDNN v7.0.5 (Dec 5, 2017), for CUDA 9.0 and follow the instruciton on the manual book.
This is what I got on my device,I am rly new on this,I rly dont know what to do.can someone help?

th> require ‘cudnn’
Found Environment variable CUDNN_PATH = /opt/cuda/lib64/libcudnn.so.5/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: /home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: /home/suice07/torch/install/share/lua/5.1/cudnn/ffi.lua:1592: /opt/cuda/lib64/libcudnn.so.5: cannot open shared object file: No such file or directory
stack traceback:
[C]: in function ‘error’
/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: in function ‘require’
[string “_RESULT={require ‘cudnn’}”]:1: in main chunk
[C]: in function ‘xpcall’
/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:661: in function ‘repl’
…ce07/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:204: in main chunk
[C]: at 0x55bd3c47a520
[0.3390s]
th> exit
Do you really want to exit ([y]/n)? y
[suice07@suice07-pc Downloads] export CUDNN_PATH="/opt/cuda/lib64/libcudnn.so.7" [suice07@suice07-pc Downloads] th

______ __ | Torch7
/_ / ________/ / | Scientific computing for Lua.
/ / / _ / __/ / _ \ | Type ? for help
/_/ _
// _//// | https://github.com/torch
| http://torch.ch

th> require ‘cudnn’
Found Environment variable CUDNN_PATH = /opt/cuda/lib64/libcudnn.so.7/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: /home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: /home/suice07/torch/install/share/lua/5.1/cudnn/ffi.lua:1618: These bindings are for CUDNN 5.x (5005 <= cudnn.version > 6000) , while the loaded CuDNN is version: 7005
Are you using an older or newer version of CuDNN?
stack traceback:
[C]: in function ‘error’
/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:389: in function ‘require’
[string “_RESULT={require ‘cudnn’}”]:1: in main chunk
[C]: in function ‘xpcall’
/home/suice07/torch/install/share/lua/5.1/trepl/init.lua:661: in function ‘repl’
…ce07/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:204: in main chunk
[C]: at 0x558a71914520
[0.3587s]

You’ve installed CUDNN 7:

“I download cuDNN v7.0.5 (Dec 5, 2017)”

but your torch/lua is looking for CUDNN 5:

/opt/cuda/lib64/libcudnn.so.5: cannot open shared object file: No such file or directory

These bindings are for CUDNN 5.x (5005 <= cudnn.version > 6000) , while the loaded CuDNN is version: 7005

You can get older versions of CUDNN from the same place that you got CUDNN 7.

You would need to install it in such a way that libcudnn.so.5 shows up at /opt/cuda/lib64

to satisfy your current install of torch/lua

the problem is ,the libcudnn.so.5 do exist in that folder.but this error still comes out.

That’s because your LD_LIBRARY_PATH or CUDNN_PATH is causing torch to find your CUDNN 7 install, instead of CUDNN 5.

export CUDNN_PATH="/opt/cuda/lib64/libcudnn.so.7"

If libcudnn.so.5 is in that location, try this:

export CUDNN_PATH="/opt/cuda/lib64/"

doesnot work.
it says.cant read the file.is a directory

OK then try this:

export CUDNN_PATH="/opt/cuda/lib64/libcudnn.so.5"

already tried.i think i will try the libcudnn.so.5 from other version