Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Issue]: Despite adding a new INFO log or modifying an existing INFO in the source code, recompiling and setting export NCCL_DEBUG=INFO, the expected log outputs do not appear #1436

Open
Kyrienn opened this issue Nov 27, 2024 · 1 comment

Comments

@Kyrienn
Copy link

Kyrienn commented Nov 27, 2024

Problem Description

When I create a new log using INFO in the source code, then compile it using ./install.sh, and test it with ./all_reduce_perf, while setting the environment variable as export NCCL_DEBUG=INFO, I cannot see the log output that I added. Moreover, I tried modifying the string in the existing INFO log at line 57 of bootstrap.cc (e.g., INFO(NCCL_INIT, "Bootstrap: Using")), recompiled as described above and executed the test again, but the INFO output still did not change. I also tried deleting the build directory, modifying the source code, and recompiling, but it still doesn't work.

Operating System

ubuntu-24.04

CPU

Intel(R) Core(TM) i7-14700k

GPU

2x AMD Radeon RX GPU 7900XT

ROCm Version

ROCm 6.2.3

ROCm Component

No response

Steps to Reproduce

No response

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

No response

Additional Information

No response

@nileshnegi
Copy link
Collaborator

nileshnegi commented Nov 27, 2024

how are you building RCCL-Tests?
can you check if ldd ./all_reduce_perf links to the correct RCCL (that you built) instead of a pre-built RCCL like under /opt/rocm/lib/librccl.so?

alternatively, you can re-try by setting LD_LIBRARY_PATH=<path-to-your-RCCL>/build/release NCCL_DEBUG=INFO ./all_reduce_perf

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants