File: cudnn_rnn_determinism.rst

package info (click to toggle)
pytorch 1.7.1-7
  • links: PTS, VCS
  • area: main
  • in suites: bullseye
  • size: 80,340 kB
  • sloc: cpp: 670,830; python: 343,991; ansic: 67,845; asm: 5,503; sh: 2,924; java: 2,888; xml: 266; makefile: 244; ruby: 148; yacc: 144; objc: 51; lex: 44
file content (16 lines) | stat: -rw-r--r-- 645 bytes parent folder | download
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
.. warning::
    There are known non-determinism issues for RNN functions on some versions of cuDNN and CUDA.
    You can enforce deterministic behavior by setting the following environment variables:
    
    On CUDA 10.1, set environment variable ``CUDA_LAUNCH_BLOCKING=1``.
    This may affect performance.

    On CUDA 10.2 or later, set environment variable
    (note the leading colon symbol)
    ``CUBLAS_WORKSPACE_CONFIG=:16:8``
    or
    ``CUBLAS_WORKSPACE_CONFIG=:4096:2``.

    See the `cuDNN 8 Release Notes`_ for more information.

.. _cuDNN 8 Release Notes: https://docs.nvidia.com/deeplearning/sdk/cudnn-release-notes/rel_8.html