8000 torch.accelerator.is_available() raise RuntimeError if no available CUDA/XPU devices · Issue #144567 · pytorch/pytorch · GitHub
[go: up one dir, main page]
Skip to content

torch.accelerator.is_available() raise RuntimeError if no available CUDA/XPU devices < 8000 span class="f1-light color-fg-muted">#144567

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
guangyey opened this issue Jan 10, 2025 · 5 comments
Closed

torch.accelerator.is_available() raise RuntimeError if no available CUDA/XPU devices #144567

guangyey opened this issue Jan 10, 2025 · 5 comments
Labels
bug high priority module: accelerator Issues related to the shared accelerator API module: regression It used to work, and now it doesn't triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Milestone
2.7.0

Comments

@guangyey
Copy link
Collaborator
guangyey commented Jan 10, 2025 •
edited by pytorch-bot bot
Loading

🐛 Describe the bug

>>> import torch
>>> torch.accelerator.is_available()
/home/guangyey/repos/stock-pytorch/torch/xpu/__init__.py:120: UserWarning: XPU device count is zero! (Triggered internally at /home/guangyey/repos/stock-pytorch/c10/xpu/XPUFunctions.cpp:117.)
  torch._C._xpu_init()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/guangyey/repos/stock-pytorch/torch/accelerator/__init__.py", line 46, in is_available
    return device_count() > 0
  File "/home/guangyey/repos/stock-pytorch/torch/accelerator/__init__.py", line 33, in device_count
    return torch._C._accelerator_deviceCount()
  File "/home/guangyey/repos/stock-pytorch/torch/xpu/__init__.py", line 120, in _lazy_init
    torch._C._xpu_init()
RuntimeError: No XPU devices are available.

The root cause is that #144368 changed the current accelerator detection from runtime to compile time. The call stack now follows this flow torch.accelerator.device_count -> device_lazy_init -> lazyInitDevice -> device_count_ensure_non_zero

As a result, a RuntimeError is raised if a user runs a PyTorch wheel built with XPU on a machine without any available XPU devices. The same issue applies to CUDA as well.

Versions

Collecting environment information...
PyTorch version: 2.7.0a0+gitcfd08f8
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.2 LTS (x86_64)
GCC version: (Ubuntu 11.3.0-1ubuntu1~22.04.1) 11.3.0
Clang version: Could not collect
CMake version: version 3.31.1
Libc version: glibc-2.35

Python version: 3.10.14 | packaged by conda-forge | (main, Mar 20 2024, 12:45:18) [GCC 12.3.0] (64-bit runtime)
Python platform: Linux-5.19.0-32-generic-x86_64-with-glibc2.35
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 42 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 24
On-line CPU(s) list: 0-23
Vendor ID: GenuineIntel
Model name: 12th Gen Intel(R) Core(TM) i9-12900
CPU family: 6
Model: 151
Thread(s) per core: 2
Core(s) per socket: 16
Socket(s): 1
Stepping: 2
CPU max MHz: 5100.0000
CPU min MHz: 800.0000
BogoMIPS: 4838.40
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req hfi umip pku ospke waitpkg gfni vaes vpclmulqdq tme rdpid movdiri movdir64b fsrm md_clear serialize pconfig arch_lbr ibt flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 640 KiB (16 instances)
L1i cache: 768 KiB (16 instances)
L2 cache: 14 MiB (10 instances)
L3 cache: 30 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-23
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected

Versions of relevant libraries:
[pip3] flake8==6.1.0
[pip3] flake8-bugbear==23.3.23
[pip3] flake8-comprehensions==3.15.0
[pip3] flake8-executable==2.1.3
[pip3] flake8-logging-format==0.9.0
[pip3] flake8-pyi==23.3.1
[pip3] flake8-simplify==0.19.3
[pip3] mypy==1.13.0
[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.26.4
[pip3] optree==0.13.0
[pip3] torch==2.7.0a0+gitcfd08f8
[conda] numpy 1.26.4 pypi_0 pypi
[conda] optree 0.13.0 pypi_0 pypi
[conda] torch 2.7.0a0+gitcfd08f8 dev_0
[conda] torchfix 0.4.0 pypi_0 pypi

cc @ezyang @gchanan @zou3519 @kadeng @msaroufim @albanD @EikanWang
@guangyey guangyey added module: accelerator Issues related to the shared accelerator API bug labels Jan 10, 2025
@malfet malfet added module: regression It used to work, and now it doesn't high priority labels Jan 10, 2025
@malfet malfet added this to the 2.6.0 milestone Jan 10, 2025
@malfet
Copy link
Contributor
malfet commented Jan 10, 2025 •
edited
Loading

[edit] I see that accelerator is a new feature in 2.6...

@kit1980 kit1980 mentioned this issue Jan 10, 2025
Closed
@huydhn
Copy link
Contributor
huydhn commented Jan 10, 2025

The failure is caught by nogpu periodic tests:

GH job link HUD commit link

They are periodic tests though , so the error surfaces few hours after the change landed https://hud.pytorch.org/hud/pytorch/pytorch/main/1?per_page=50&name_filter=nogpu&mergeLF=true

@guangyey
Copy link
Collaborator Author

@albanD torch.accelerator.device_count has some subtle differences compared with torch.xxx.device_count. For example, torch.cuda.device_count can be called before forking, as it does not initiate CUDA runtime. Another difference is that torch.cuda.device_count can parse environment variables CUDA_VISIBLE_DEVICES.

Do you think we should implement torch.accelerator.device_count using PyImport_ImportModule(module) and PyObject_CallMethod(module, "device_count")? This would be equivalent to calling torch.get_device_module().device_count, but implemented in C++, helping to bridge the gap between the two methods. This design could eliminate this RuntimeError issue as well.

Or I will find a way to replace this line of code to fix this issue. What do you think about it?

@guangyey guangyey mentioned this issue Jan 13, 2025
Closed
@atalman
Copy link
Contributor
atalman commented Jan 13, 2025 •
edited
Loading

Looks like rc for 2.6 is not affected:

CUDA 12.4, on GPU machine

>>> import torch
>>> torch.__version__
'2.6.0+cu124'
>>> torch.accelerator.is_available()
True
>>> exit()

CPU, on GPU machine:

>>> import torch
>>> torch.__version__
'2.6.0+cpu'
>>> torch.accelerator.is_available()
False

CUDA 12.4 on CPU machine:

>>> import torch
>>> torch.__version__
'2.6.0+cu124'
>>> torch.accelerator.is_available()
False

@jbschlosser jbschlosser removed this from the 2.6.0 milestone Jan 13, 2025
@jbschlosser jbschlosser added triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module and removed triage review labels Jan 13, 2025
@guangyey guangyey mentioned this issue Jan 15, 2025
Closed
@atalman atalman added this to the 2.6.1 milestone Jan 21, 2025
@atalman atalman modified the milestones: 2.6.1, 2.7.0 Mar 4, 2025
@guangyey
Copy link
Collaborator Author
guangyey commented Mar 5, 2025

Closed as the PR introduces this regression is reverted.

@guangyey guangyey closed this as completed Mar 5, 2025
@atalman atalman mentioned this issue Apr 3, 2025
Closed
65 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
bug high priority module: accelerator Issues related to the shared accelerator API module: regression It used to work, and now it doesn't triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Projects
None yet
Milestone
2.7.0
Development

No branches or pull requests
5 participants
@huydhn @malfet @atalman @jbschlosser @guangyey
0