Skip to content

torch-2.0.0-rc1 and torch-1.13.1 can not be installed on Ubuntu 20.04  #91067

@malfet

Description

@malfet

🐛 Describe the bug

  1. Allocate c5a.4xlarge instance, for example by running:
import boto3
ec2=boto3.resource("ec2")
rc=ec2.create_instances(ImageId="ami-031843d9eaa76ad7a",InstanceType="c5a.4xlarge",SecurityGroups=['ssh-allworld'],KeyName="nshulga-key",MinCount=1,MaxCount=1,BlockDeviceMappings=[{'DeviceName': '/dev/sda1','Ebs': {'DeleteOnTermination': True, 'VolumeSize': 150,'VolumeType': 'standard'}}])
  1. SSH into the instance and run python3 -mpip install torch
  2. Run python3 -c "import torch"

Above fails with:

$ python3 -c "import torch"
Traceback (most recent call last):
  File "/home/ubuntu/.local/lib/python3.8/site-packages/torch/__init__.py", line 172, in _load_global_deps
    ctypes.CDLL(lib_path, mode=ctypes.RTLD_GLOBAL)
  File "/usr/lib/python3.8/ctypes/__init__.py", line 373, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: /home/ubuntu/.local/lib/python3.8/site-packages/torch/lib/../../nvidia/cublas/lib/libcublas.so.11: undefined symbol: cublasLtHSHMatmulAlgoInit, version libcublasLt.so.11

Versions

1.13.1, 1.13.0

cc @ezyang @gchanan @zou3519 @seemethere

Metadata

Metadata

Labels

high prioritymodule: binariesAnything related to official binaries that we release to userstriagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions