Installation on Ubuntu 20.04
Hello,
I am trying to install softgym on 2004. My cuda version is 11.4 and my driver is 470.129. I am using an RTX 3060 card.
I have successfully compiled Pyflex using the command in prepare_1.0.sh and compile_1.0.sh. However, when I am trying to run the example, I got the following error:
Traceback (most recent call last):
File "examples/random_env.py", line 5, in <module>
from softgym.registered_env import env_arg_dict, SOFTGYM_ENVS
File "/home/hongtao/src/softgym/softgym/registered_env.py", line 1, in <module>
from softgym.envs.pour_water import PourWaterPosControlEnv
File "/home/hongtao/src/softgym/softgym/envs/pour_water.py", line 4, in <module>
import pyflex
ImportError: /home/hongtao/src/softgym/PyFlex/bindings/build/pyflex.cpython-38-x86_64-linux-gnu.so: undefined symbol: cudaSetupArgument
I checked that one reason for this problem is that the running cuda version is not matched with the compiled one (#1705). But my compiling cuda version and the one shown from nvcc -V are both 11.4.
Is there a solution to this problem? Thanks!
maybe you can try docker?
Hello,
I am trying to install softgym on 2004. My cuda version is 11.4 and my driver is 470.129. I am using an RTX 3060 card.
I have successfully compiled Pyflex using the command in
prepare_1.0.shandcompile_1.0.sh. However, when I am trying to run the example, I got the following error:Traceback (most recent call last): File "examples/random_env.py", line 5, in <module> from softgym.registered_env import env_arg_dict, SOFTGYM_ENVS File "/home/hongtao/src/softgym/softgym/registered_env.py", line 1, in <module> from softgym.envs.pour_water import PourWaterPosControlEnv File "/home/hongtao/src/softgym/softgym/envs/pour_water.py", line 4, in <module> import pyflex ImportError: /home/hongtao/src/softgym/PyFlex/bindings/build/pyflex.cpython-38-x86_64-linux-gnu.so: undefined symbol: cudaSetupArgumentI checked that one reason for this problem is that the running cuda version is not matched with the compiled one (#1705). But my compiling cuda version and the one shown from
nvcc -Vare both 11.4.Is there a solution to this problem? Thanks!
I have the same problem as you. How did you solve it?