GPU error when running node worker
jrieffel opened this issue · comments
I get the following printed to a history file when running vx3_node_worker. I have an NVIDIA Quadro K1200 (rev a2), and I'm using the custom CMake file for my GPU you provided in a separate thread.
1 GPU found.
=== set device to 0 for 1 simulations ===
Total GPU memory 4236902400 bytes.
Set GPU heap size to be 2118451200 bytes.
/home/rieffelj/src/voxcraft-sim/src/VX3/VX3_SimulationManager.cu(415): CUDA Func
tion Error: too many resources requested for launch
Running simulation locally by default.
./vx3_node_worker -i workspace/locally/20200731155914.vxt -o workspace/locally/2
0200731155914.vxr
it says CUDA Func
tion Error: too many resources requested for launch.
i guess it's the reason.
Thanks for all your hard work, voxcraft team! I think I was able to resolve the issue by booting my research computer into single user mode, thereby freeing up the GPU. Worth keeping this in mind for future resource-poor researchers!
Meanwhile, saving up for a Tesla..
@jrieffel if you document your process, we will share it as "voxcraft for resource-poor researchers" documentation.
Either way, we are looking forward to see how your lab uses voxcraft.
Keep the issues/pull requests/questions coming!
Sam