nvshmemx_getmem_nbi_block failed when test nvshmem in H800

when i compiled from source, it success. But when I run: bash ./launch.sh ./python/triton_dist/test/nvidia/test_nvshmem_api.py to test nvshmem, happend:

when I compile, something happend:

I'm wondering if there is a connection between these two.

And when I run: bash ./launch.sh ./python/triton_dist/test/nvidia/test_gemm_rs.py 8192 8192 29568，the output is:

Reproduce the reported commands on an H800: bash ./launch.sh ./python/triton_dist/test/nvidia/test_nvshmem_api.py and test_gemm_rs.py with the given dimensions. Inspect the build and test output shown in the issue, then trace the failing nvshmem test to determine whether the source-build behavior is related. Done means the test completes successfully or the failure is narrowed to a documented, reproducible cause.

Written by the indexing model from the issue text.