PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!

Autonomous Machines Robotics - Isaac Isaac Gym
1 
rlgpu ❯ python train.py --task Cartpole
Importing module 'gym_37' (/home/lqk/downloads/IsaacGym_Preview_2_Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/gym_37.so)
Setting GYM_USD_PLUG_INFO_PATH to /home/lqk/downloads/IsaacGym_Preview_2_Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/usd/plugInfo.json
PyTorch version 1.8.1
Device count 1
/home/lqk/downloads/IsaacGym_Preview_2_Package/isaacgym/python/isaacgym/_bindings/src/gymtorch
Using /home/lqk/.cache/torch_extensions as PyTorch extensions root...
Emitting ninja build file /home/lqk/.cache/torch_extensions/gymtorch/build.ninja...
Building extension module gymtorch...
Using envvar MAX_JOBS (10) as the number of workers...
ninja: no work to do.
Loading extension module gymtorch...
Importing module 'rlgpu_37' (/home/lqk/downloads/IsaacGym_Preview_2_Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/rlgpu_37.so)
Setting seed: 8963
Python
Not connected to PVD
+++ Using GPU PhysX
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 134217728 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 402653184 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 100663296 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
PxgCudaDeviceMemoryAllocator fail to allocate memory 67108864 bytes!! Result = 2
Physics Engine: PhysX
Physics Device: cuda:0
GPU Pipeline: enabled
^C^\[1]    19158 quit (core dumped)  python train.py --task Cartpole

I have checked vulkaninfo and used export VK_ICD_FILENAMES=/usr/share/vulkan/icd.d/nvidia_icd.json.
The joint_monkey example works fine with CPU. The graphics card driver is 470.63.01.

But I am using a pretty old graphics card: GTX750, not sure if this caused the problem.

Thanks to kind people!

2

I tested Isacc Gym in my colleague’s computer, which has GTX1650, the examples run correctly

3

We haven’t really tested on those GPUs, but maybe you could try running the example with less number of environments and see if that helps? python train.py --task Cartpole --num_envs 10 for example.