CUDA 9.2 RNN non-determinism

1

In the LSTM page, I see the 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.

I use CUDA 9.2, so, for CUDA 9.2 what should I do to enforce deterministic behavior? I set the random seed as follows:

    torch.manual_seed(seed)
    np.random.seed(seed)
    random.seed(seed)
    torch.backends.cudnn.deterministic = True
    torch.backends.cudnn.benchmark = False

Thanks.

2

Are you seeing the non-determinisim also when disabling cudnn via torch.backends.cudnn.enabled = False? If not, which cudnn version are you using?