You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, recently we received an issue from community users like this:
File "/home/xinanlin/xinanlin/pytorch-main/torch/utils/_triton.py", line 111, in triton_hash_with_backend
backend = triton_backend()
File "/home/xinanlin/xinanlin/pytorch-main/torch/utils/_triton.py", line 103, in triton_backend
target = driver.active.get_current_target()
File "/home/xinanlin/xinanlin/miniforge3/envs/torch-py3.9/lib/python3.9/site-packages/triton/backends/intel/driver.py", line 553, in get_current_target
device = self.get_current_device()
File "/home/xinanlin/xinanlin/miniforge3/envs/torch-py3.9/lib/python3.9/site-packages/triton/backends/intel/driver.py", line 545, in get_current_device
return self.utils.get_current_device()
File "/home/xinanlin/xinanlin/miniforge3/envs/torch-py3.9/lib/python3.9/site-packages/triton/backends/intel/driver.py", line 539, in __getattr__
self.utils = XPUUtils()
File "/home/xinanlin/xinanlin/miniforge3/envs/torch-py3.9/lib/python3.9/site-packages/triton/backends/intel/driver.py", line 154, in __init__
mod = compile_module_from_src(Path(os.path.join(dirname, "driver.c")).read_text(), "spirv_utils")
File "/home/xinanlin/xinanlin/miniforge3/envs/torch-py3.9/lib/python3.9/site-packages/triton/backends/intel/driver.py", line 135, in compile_module_from_src
mod = importlib.util.module_from_spec(spec)
File "<frozen importlib._bootstrap>", line 565, in module_from_spec
File "<frozen importlib._bootstrap_external>", line 1173, in create_module
File "<frozen importlib._bootstrap>", line 228, in _call_with_frames_removed
torch._inductor.exc.InductorError: ImportError: libsycl.so.7: cannot open shared object file: No such file or directory
The root cause is:
Community users used the old tool chain pytorch-gpu-dev-0.5( libsycl.so.7) and cached the spirv_utils.so in ~/.triton/cache
Then after some days, they upgrade the tool chain to oneapi 2025(libsycl.so.8) but the cache is not cleaned, then they got the above error. We've asked the users delete the cache to resolve the error.
But I think we shouldn't allow this error to be exposed to users.
Environment details
Triton release 3.2
The text was updated successfully, but these errors were encountered:
Describe the bug
Hi, recently we received an issue from community users like this:
The root cause is:
Community users used the old tool chain pytorch-gpu-dev-0.5( libsycl.so.7) and cached the spirv_utils.so in ~/.triton/cache
Then after some days, they upgrade the tool chain to oneapi 2025(libsycl.so.8) but the cache is not cleaned, then they got the above error. We've asked the users delete the cache to resolve the error.
But I think we shouldn't allow this error to be exposed to users.
Environment details
Triton release 3.2
The text was updated successfully, but these errors were encountered: