kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-14 06:27:45 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

[https://nvbugs/5501820][fix] Add requirements for numba-cuda version to WAR mem corruption (#7992) (#8414)

Browse Source 
Signed-off-by: Pengbo Wang <221450789+pengbowang-nv@users.noreply.github.com>
This commit is contained in:
Pengbo Wang 2025-10-20 15:01:08 +08:00 committed by GitHub
parent 14d0f5d683
commit 8ce2dc5cb7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
requirements.txt
View File

@ -76,3 +76,5 @@ blobfile
openai-harmony==0.0.4
nvidia-cutlass-dsl==4.2.1; python_version >= "3.10"
numexpr<2.14.0 # WAR for attempted use of nonexistent numpy.typing
# numba-cuda # <For CUDA 12.9>
numba-cuda>=0.19.0 # WAR for nvbugs/5501820

2
tests/unittest/_torch/multi_gpu/test_linear.py
View File

@ -22,8 +22,6 @@ MPI.pickle.__init__(
# needed since we reuse the mpi executor pool, first test running will leak a thread
pytestmark = pytest.mark.threadleak(enabled=False)
pytest.skip(reason="https://nvbugspro.nvidia.com/bug/5501820",
allow_module_level=True)
def rms_norm(x: torch.Tensor, weight: torch.Tensor = None, eps: float = 1e-6):