Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

UnboundLocalError on cuda/memory.pyx

See original GitHub issue

Hi!

I’m having an error when using a GPU from SpaCy, but originated in a cupy module. The relevant (and repeatable) traceback is as follows:

 File "spacy/ml/parser_model.pyx", line 327, in spacy.ml.parser_model.step_forward.backprop_parser_step
  File "spacy/ml/parser_model.pyx", line 277, in spacy.ml.parser_model.ParserStepModel.backprop_step
  File "/home/user/.cache/pypoetry/virtualenvs/notebooks-noypvBLr-py3.8/lib/python3.8/site-packages/spacy/util.py", line 958, in get_async
    array.set(numpy_array, stream=stream)
  File "cupy/core/core.pyx", line 1622, in cupy.core.core.ndarray.set
  File "cupy/core/core.pyx", line 1649, in cupy.core.core.ndarray.set
  File "cupy/core/core.pyx", line 1651, in cupy.core.core.ndarray.set
  File "cupy/cuda/memory.pyx", line 479, in cupy.cuda.memory.MemoryPointer.copy_from_host_async
UnboundLocalError: local variable 'ptr' referenced before assignment

I think the solution is pretty straightforward. In these lines https://github.com/cupy/cupy/blob/2e95ed7e994b1db230501a69e8a323666164fb80/cupy/cuda/memory.pyx#L462-L480 the variable ptr is effectively undefined when the stream is not None. I’m not very used to the library, but I think that just by swapping L473 with L474 so that ptr is defined regardless of the value of stream, it should get sorted out.

Let me know if I can help in any way 😃

Issue Analytics

State:
Created 3 years ago
Comments:8 (8 by maintainers)

Top GitHub Comments

2reactions

viantirreaucommented, Mar 17, 2021

Of course 😃 I’ll send it right away. Just arrived home, I’m currently on v9.0.0b3

1reaction

viantirreaucommented, Mar 17, 2021

Thanks for your fast response, @leofang I’ll try to build a minimal example from SpaCy’s code

Top Results From Across the Web

python - Cupy memory error on Google Colab with GPU

_malloc() cupy/cuda/memory.pyx in cupy.cuda.memory._try_malloc() OutOfMemoryError: Out of memory allocating 7,200,000,000 bytes (allocated ...

NVIDIA CUDA cannot work properly

Hello guys: I try to run a machine learning program with NVIDIA RTX A6000 graphical cards and Pytorch, but confronted with the following ......

Frequently Asked Questions — PyTorch 1.13 documentation

Frequently Asked Questions. My model reports “cuda runtime error(2): out of memory”. As the error message suggests, you have run out of memory...

cupy.cuda.memory.OutOfMemoryError problem - ner

Hi, I am trying to train with an NER model using en_core_web_trf base model and getting this error! (venv) C:\Users\Asli>python -m prodigy ...

chainer/develop-cupy - Gitter

_malloc() cupy/cuda/memory.pyx in cupy.cuda.memory._try_malloc() OutOfMemoryError: out of memory to allocate 5120000000 bytes (total 20480000000 bytes).