search

lightvector / KataGo

GTP engine and self-play learning in Go
https://katagotraining.org/
Other
3.6k stars 568 forks source link

Katago on Linux, CUDA backend, throws StringError exception #1002

Open emogenet opened 4 days ago

emogenet commented 4 days ago

KataGo v1.15.3 Using TrompTaylor rules initially, unless GTP/GUI overrides this terminate called after throwing an instance of 'StringError' what(): CUDA Error, for createComputeHandle file /home/emogenet/Desktop/KataGo/cpp/neuralnet/cudabackend.cpp, func cudaSetDevice(gpuIdxForThisThread), line 2405, error unknown error Aborted (core dumped)

yehu3d commented 4 days ago

maybe you have 1 gpu but you set use 2 gpus

emogenet commented 4 days ago

maybe you have 1 gpu but you set use 2 gpus

No, I am on a system with Intel i915 + NVidia GeForce RTX2050

I debugged katago, and it looks like that cryptic message is Katago can't find a proper CUDA device, even though I can run a 3B LLM on it.