Problem with packaging and distributing ChronoEngine_sensor.dll

50 views
Skip to first unread message

Dinesh M

unread,
Mar 10, 2023, 2:54:24 PM3/10/23
to projec...@googlegroups.com
Hi, 

I've installed projectChrono from source in my laptop with T1200 GPU and linked the dlls from install directory to an external project. My project looks just the same as demo_SEN_lidar program. While installation I've also enabled CUDA NVRTC in cmake configuration. 

Everything works fine with the laptop. I'm able to see the LIDAR point cloud visualization. 

Now when I use the same dll (copy paste the chronoEngineSensor.dll and lib and .cu files from the install folder) to my workstation PC(with A4000 GPU), the shader gets compiled fine and no error is thrown but the LIDAR and Camera sensor data and visualization comes out to be blank.

Only if I build the dll again in workstation seperately. It works. I'm kinda looking for a way where I simply export a dll and it works across every other systems and not needing to build the sensor dll each time for different PCs. 

How can I achieve this and what am I missing?

Thanks,
Dinesh

Radu Serban

unread,
Mar 29, 2023, 12:24:10 PM3/29/23
to Dinesh M, projec...@googlegroups.com

Dinesh,

 

I have not tried this myself, but you should try to force CUDA_ARCH_NAME=All during CMake configuration on whatever machine you want to build the code on in order to generate cuda for a list of architectures and not only the one on the build machine (see FindCudaArch.cmake in the Chrono distribution)

 

--Radu

--
You received this message because you are subscribed to the Google Groups "ProjectChrono" group.
To unsubscribe from this group and stop receiving emails from it, send an email to projectchron...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/CALZuBmtmpG1QkKvNM0FeC9-0gbkgrz3M4fM0VSeMM72Y1zFnQw%40mail.gmail.com.

Dinesh M

unread,
Mar 29, 2023, 4:14:45 PM3/29/23
to Radu Serban, projec...@googlegroups.com
Hi Radu,


    Thanks for your reply. Setting the CUDA_ARCH_NAME to "All" seem to have solved my problem. !!  

Thank you ;)


Reply all
Reply to author
Forward
0 new messages