-
Epic
-
Resolution: Done
-
Critical
-
None
-
None
-
Build Torch and other packages with NumPy 2.x
-
False
-
-
False
-
Done
-
0% To Do, 0% In Progress, 100% Done
-
Proposed
We are patching some packages to force "numpy<2.0" at build time. This version cap is now causing problems. Packages like Torch do not work when they are compiled with NumPy 1.x but the environment has NumPy 2.x. The other way is supported, though.
Pinning to "numpy <2.0" globally does not work, because "numba >= 0.60" requires "numpy >= 2.0" to build. "numba < 0.60" requires an older "llvmlite" to build, which in return requires LLVM 14. Our builder image now has LLVM 15. We need to rebuild Torch and other packages with NumPy 2.x.
I think RHELAI is getting lucky, because we also patch xformers to have an installation requirement on "numpy < 2.0".
A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.2.5 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.
Traceback (most recent call last): File "/opt/app-root/bin/vllm", line 5, in <module>
from vllm.entrypoints.cli.main import main
File "/opt/app-root/lib64/python3.11/site-packages/vllm/__init__.py", line 10, in <module>
import vllm.env_override # isort:skip # noqa: F401
File "/opt/app-root/lib64/python3.11/site-packages/vllm/env_override.py", line 4, in <module>
import torch
File "/opt/app-root/lib64/python3.11/site-packages/torch/__init__.py", line 2222, in <module>
from torch import quantization as quantization # usort: skip
File "/opt/app-root/lib64/python3.11/site-packages/torch/quantization/__init__.py", line 2, in <module>
from .fake_quantize import * # noqa: F403
File "/opt/app-root/lib64/python3.11/site-packages/torch/quantization/fake_quantize.py", line 10, in <module>
from torch.ao.quantization.fake_quantize import (
File "/opt/app-root/lib64/python3.11/site-packages/torch/ao/quantization/__init__.py", line 12, in <module>
from .pt2e._numeric_debugger import ( # noqa: F401
File "/opt/app-root/lib64/python3.11/site-packages/torch/ao/quantization/pt2e/_numeric_debugger.py", line 9, in <module>
from torch.export import ExportedProgram
File "/opt/app-root/lib64/python3.11/site-packages/torch/export/__init__.py", line 68, in <module>
from .decomp_utils import CustomDecompTable
File "/opt/app-root/lib64/python3.11/site-packages/torch/export/decomp_utils.py", line 5, in <module>
from torch._export.utils import (
File "/opt/app-root/lib64/python3.11/site-packages/torch/_export/__init__.py", line 48, in <module>
from .wrappers import _wrap_submodules
File "/opt/app-root/lib64/python3.11/site-packages/torch/_export/wrappers.py", line 7, in <module>
from torch._higher_order_ops.strict_mode import strict_mode
File "/opt/app-root/lib64/python3.11/site-packages/torch/_higher_order_ops/__init__.py", line 1, in <module>
from torch._higher_order_ops.cond import cond
File "/opt/app-root/lib64/python3.11/site-packages/torch/_higher_order_ops/cond.py", line 9, in <module>
import torch._subclasses.functional_tensor
File "/opt/app-root/lib64/python3.11/site-packages/torch/_subclasses/functional_tensor.py", line 45, in <module>
class FunctionalTensor(torch.Tensor):
File "/opt/app-root/lib64/python3.11/site-packages/torch/_subclasses/functional_tensor.py", line 275, in FunctionalTensor
cpu = _conversion_method_template(device=torch.device("cpu"))
/opt/app-root/lib64/python3.11/site-packages/torch/_subclasses/functional_tensor.py:275: UserWarning: Failed to initialize NumPy:
A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.2.5 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.
- is related to
-
-
- New
-
