cuda-gdb not working error code
/usr/local/cuda-12.5/bin/cuda-gdb-python3.12-tui: symbol lookup error: /usr/local/cuda-12.5/bin/cuda-gdb-python3.12-tui: undefined symbol: curscr
NVIDIA CUDA 12.5 , Linux Driver 555.52.04
, Kernel: Linux tumbleweed 6.10.0-rc5-1.g0c5f39a-default #1 SMP PREEMPT_DYNAMIC Sun Jun 23 21:37:17 UTC 2024 (0c5f39a) x86_64 x86_64 x86_64 GNU/Linux, whats the reason , how can i fix it
Thanks for reaching out. How are you starting cuda-gdb in this case? Are you trying to invoke /usr/local/cuda-12.5/bin/cuda-gdb-python3.12-tui directly?
Hi
i started the CUDA_GDB as follows : cuda-gdb --version
now i have installed the CUDA 12.5 Update 1 the same Error
Without a working CUDA-GDB debugger, the NVIDIA CUDA 12.5 toolkit is worthless and unusable
regards
Franz
the invoke of directly of cuda-gdb-python3.12.tui produce the same error
As a workaround, you can try explicitly starting:
/usr/local/cuda-12.5/bin/cuda-gdb-minimal
This variant lacks python and TUI integration. We will fix the decision logic in a future CUDA release. When fixed, the workaround will no longer be necessary.
@fbernasek535 can you also provide us with the output from:
ldd /usr/local/cuda-12.5/bin/cuda-gdb-python3.12-tui
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.