erwincoumans / tiny-differentiable-simulator

Tiny Differentiable Simulator is a header-only C++ and CUDA physics library for reinforcement learning and robotics with zero dependencies.
Apache License 2.0
1.24k stars 130 forks source link

Where can I get "cuda_model_laikago.dll" ? #210

Open hjyu94 opened 1 year ago

hjyu94 commented 1 year ago

notfound

Error

Version = 4.6.0 NVIDIA 531.14
Vendor = NVIDIA Corporation
Renderer = NVIDIA GeForce RTX 3090/PCIe/SSE2
Loading URDF "C:\Users\GLAB\Downloads\dev\tiny-differentiable-simulator\data\plane_implicit.urdf".
Loading URDF "C:\Users\GLAB\Downloads\dev\tiny-differentiable-simulator\data\laikago\laikago_toes_zup_xyz_xyzrot.urdf".
v.geom_type=3v.geom_type=3v.geom_type=3v.geom_type=3v.geom_type=0v.geom_type=3v.geom_type=3v.geom_type=3v.geom_type=0v.geom_type=3v.geom_type=3v.geom_type=3v.geom_type=0v.geom_type=3v.geom_type=3v.geom_type=3v.geom_type=0Creating VectorizedEnvironment for laikago
Creating a normal distribution of size 25000000.
Finished creating a normal distribution.
Could not find file "cuda_model_laikago.dll".
Assertion failed: found, file C:\Users\GLAB\Downloads\dev\tiny-differentiable-simulator\examples\ars\ars_train_policy_cuda.cpp, line 333

After running ars_train_policy_cuda project, I got this error. Where can I get "cuda_model_laikago.dll" file?

erwincoumans commented 1 year ago

It has been a while that I looked at that code. You need to build and run another executable (that uses CppAD/CppADCodeGen) that will generate the CUDA DLL, and then make sure that DLL is in the right location (I used symbolic links for that).

This CUDA code generation is very experimental and not user friendly. I highly recommend looking into NVIDIA Warp, Google Brax/Jax or OpenAL Triton, DiffTaichi or Nimble Physics instead (unless you have time digging into this C++ code).

hjyu94 commented 1 year ago

I truly appreciate your time and effort in helping me with this matter. Your expertise and insights in this field have been invaluable to me, and I cannot thank you enough for your continuous support.

Sincerely

2023년 5월 1일 (월) 오후 2:06, erwincoumans @.***>님이 작성:

It has been a while that I looked at that code. You need to build and run another executable (that uses CppAD/CppADCodeGen) that will generate the CUDA DLL, and then make sure that DLL is in the right location (I used symbolic links for that).

It is not user friendly. I highly recommend looking into NVIDIA Warp, Google Brax/Jax or OpenAL Triton instead (unless you have time digging into this C++ code).

— Reply to this email directly, view it on GitHub https://github.com/erwincoumans/tiny-differentiable-simulator/issues/210#issuecomment-1529352613, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJWM5267Y4Y4UR56CX5TX4LXD5AEFANCNFSM6AAAAAAVYXT3C4 . You are receiving this because you authored the thread.Message ID: @.*** com>