From where to learn to use cudnn(python) within cupy? Please share some links to documentation or projects.

Thank you for using and contributing to CuPy!

If your issue is a request for support in using CuPy, please post it on Stack Overflow or Google Groups (en / ja). Developer team members are monitoring questions on these channels.

If it is a bug report, please include the following information:

• Conditions (you can just paste the output of python -c 'import cupy; cupy.show_config()')
  • CuPy version
  • OS/Platform
  • CUDA version
  • cuDNN/NCCL version (if applicable)
• Code to reproduce
• Error messages, stack traces, or logs

[emcastillo]

@leofang dude, you are the best

[leofang]

Hi @LK54345

Can you recommend any example cupy code on using texture memory?

You can take a look at how texture/surface memory is tested: https://github.com/cupy/cupy/blob/master/tests/cupy_tests/cuda_tests/test_texture.py. The tests should be close enough to real-world use cases.

Documentation is a bit tricky.

Right, admittedly it is, but it’s mainly for easy migration from CUDA C or PyCUDA to CuPy, in which users already have a validated running code utilizing texture memory. For those cases, it is straightforward to translate to our API, but there’d be a learning curve, regardless whether you’re using CUDA C, PyCUDA, or CuPy, if you are going to build the code from scratch. Texture memory works very differently, after all.

Just FYI, for these inquiries please feel free to ask on StackOverflow (and tag your post with “cupy”, which would be monitored by the team’s bot) or on CuPy’s mailing list: https://groups.google.com/forum/#!forum/cupy.