kuz/DeepMind-Atari-Deep-Q-Learner

run_gpu: module 'cutorch' not found

190n opened this issue · 13 comments

190n commented

When I try to run a game on the GPU, I get the following output:

$ ./run_gpu breakout
-framework alewrap -game_path /home/ben/Downloads/DeepMind-Atari-Deep-Q-Learner-master/roms/ -name DQN3_0_1_breakout_FULL_Y -env breakout -env_params useRGB=true -agent NeuralQLearner -agent_params lr=0.00025,ep=1,ep_end=0.1,ep_endt=replay_memory,discount=0.99,hist_len=4,learn_start=50000,replay_memory=1000000,update_freq=4,n_replay=1,network="convnet_atari3",preproc="net_downsample_2x_full_y",state_dim=7056,minibatch_size=32,rescale_r=1,ncols=1,bufferSize=512,valid_size=500,target_q=10000,clip_delta=1,min_reward=-1,max_reward=1 -steps 50000000 -eval_freq 250000 -eval_steps 125000 -prog_freq 10000 -save_freq 125000 -actrep 4 -gpu 0 -random_starts 30 -pool_frms type="max",size=2 -seed 1 -threads 4
Fontconfig error: "/home/ben/.config/font-manager/local.conf", line 2: syntax error
Gtk-Message: Failed to load module "canberra-gtk-module"
Torch Threads:  4
qlua: ./initenv.lua:58: module 'cutorch' not found:
    no field package.preload['cutorch']
    no file './cutorch.lua'
    no file '/home/ben/Downloads/DeepMind-Atari-Deep-Q-Learner-master/torch/share/luajit-2.0.3/cutorch.lua'
    no file '/usr/local/share/lua/5.1/cutorch.lua'
    no file '/usr/local/share/lua/5.1/cutorch/init.lua'
    no file '/home/ben/Downloads/DeepMind-Atari-Deep-Q-Learner-master/torch/share/lua/5.1/cutorch.lua'
    no file '/home/ben/Downloads/DeepMind-Atari-Deep-Q-Learner-master/torch/share/lua/5.1/cutorch/init.lua'
    no file './cutorch.so'
    no file '/usr/local/lib/lua/5.1/cutorch.so'
    no file '/home/ben/Downloads/DeepMind-Atari-Deep-Q-Learner-master/torch/lib/lua/5.1/cutorch.so'
    no file '/usr/local/lib/lua/5.1/loadall.so'
stack traceback:
    [C]: at 0x7f0b12f8d7b0
    [C]: in function 'require'
    ./initenv.lua:58: in function 'torchSetup'
    ./initenv.lua:112: in function 'setup'
    train_agent.lua:52: in main chunk
kuz commented

This is probably because you do not have nVidia CUDA toolkit installed.

  • 😄2
190n commented

Thanks. Trying that now...

190n commented

Still doesn't work.

snapo commented

Problem is that luarocks install cunn and luarocks install cutorch failed on your system. (Same as on mine with ubuntu 15.10 / CUDA 7.5 ) will try to find a workaround.

--- Update ---
Problem was i did not had the same version for the c compiler and for the c++ compiler.
Second problem was i the g++ symlink in /usr/bin/ was missing after installing gcc/g++ 4.9

to build the packages this did work for me:

apt-get install g++-4.9 gcc-4.9
rm -rf /usr/bin/gcc
rm -rf /usr/bin/g++
ln -s /usr/bin/g++-4.9 /usr/bin/g++
ln -s /usr/bin/gcc-4.9 /usr/bin/gcc

now execute the install dependencies.sh again

Then if you are using a server that does commandline only open the file
%yourDQN%/dqn/train_agent.lua

comment out (--) line 98 to not use X11
So that it looks like this:
-- win =image.display({image=screen, win=win})

Hope this helps you

drsstein commented

I am having the same issue despite running both g++ and gcc with the same version (4.8.2. in my case).

mhmtsarigul commented

I have the same problem in Ubuntu 14.04. Cutorch is already installed.

qlua: ./initenv.lua:58: module 'cutorch' not found:
no field package.preload['cutorch']
no file './cutorch.lua'
no file '/home/mehmet/Desktop/DeepMind-Atari-Deep-Q-Learner-master/torch/share/luajit-2.0.4/cutorch.lua'
no file '/usr/local/share/lua/5.1/cutorch.lua'
no file '/usr/local/share/lua/5.1/cutorch/init.lua'
no file '/home/mehmet/Desktop/DeepMind-Atari-Deep-Q-Learner-master/torch/share/lua/5.1/cutorch.lua'
no file '/home/mehmet/Desktop/DeepMind-Atari-Deep-Q-Learner-master/torch/share/lua/5.1/cutorch/init.lua'
no file './cutorch.so'
no file '/usr/local/lib/lua/5.1/cutorch.so'
no file '/home/mehmet/Desktop/DeepMind-Atari-Deep-Q-Learner-master/torch/lib/lua/5.1/cutorch.so'
no file '/usr/local/lib/lua/5.1/loadall.so'
stack traceback:
[C]: at 0x7feabae9f9f0
[C]: in function 'require'
./initenv.lua:58: in function 'torchSetup'
./initenv.lua:112: in function 'setup'
train_agent.lua:52: in main chunk

Cutorch is in;

cutorch scm-1 is now built and installed in /home/mehmet/torch/install/ (license: BSD)

How can i make it find cutorch?

snapo commented

Could you check your path variables (global and user) if you have added the cutorch path from your system?

mhmtsarigul commented

Thanks for your answer. I remove torch and cutorch then i install them
again maybe five times. But at the end it works.
There was a version problem i think because lua 5.1 comes with torch but
the application looks for 5.2.

2016-06-03 13:50 GMT+03:00 snapo notifications@github.com:

Could you check your path variables (global and user) if you have added
the cutorch path from your system?

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#5 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/ASxuXL1o10BTgfHiefEcW9Ed1Zm3cTGJks5qIAcOgaJpZM4F0ra-
.

jsrimr commented

@kuz I finally solved the issue passing two weeks of agony. Thanks! I think 190n didn't just follow the instruction correctly. He should have modified the path and LD_LIBRARY_PATH. 190n, you can check whether your installation completed by running samples specified in step 6-2

gaohaoyue commented

@190n Have you resolved your problem yet? Cuz I'm having the same issue.

gaohaoyue commented

@kuz What are path and LD_LIBRARY_PATH have you modified?

190n commented

@gaohaoyue I lost interest :/

Deleted user commented

I had the same problem. The following solution worked for me:

apt-get install g++-4.8 gcc-4.8
rm -rf /usr/bin/gcc
rm -rf /usr/bin/g++
ln -s /usr/bin/g++-4.8 /usr/bin/g++
ln -s /usr/bin/gcc-4.8 /usr/bin/gcc

now execute the install dependencies.sh again