cudnn 8.0 support #74

Closed
opened 2020-07-13 13:32:36 +02:00 by samsonadmin · 2 comments
samsonadmin commented 2020-07-13 13:32:36 +02:00 (Migrated from github.com)

error: ‘CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST’ was not declared in this scope
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST, 0, &bwAlgo) );

error: ‘CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST’ was not declared in this scope
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST, 0, &bwAlgo) );

AastaLLL gave some help:
https://raw.githubusercontent.com/AastaNV/JEP/master/script/topics/0001-fix-for-cudnn_v8-limited-memory-to-default-darknet-s.patch

error: ‘CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST’ was not declared in this scope CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST, 0, &bwAlgo) ); error: ‘CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST’ was not declared in this scope CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST, 0, &bwAlgo) ); AastaLLL gave some help: https://raw.githubusercontent.com/AastaNV/JEP/master/script/topics/0001-fix-for-cudnn_v8-limited-memory-to-default-darknet-s.patch
ceccocats commented 2020-07-13 23:35:04 +02:00 (Migrated from github.com)

Actually cudnn8.0.0 support was added with this commit:
https://github.com/ceccocats/tkDNN/commit/20303ac32ef8e28ca0ff94be633478d7755ae307
You are probably using 8.0.1
Nvidia changes a lot, it is not simple to keep updated.
We decided to keep updated with the versions shipped with Jetpack. The current release is:
Jetpack 4.4 (CUDA 10.2, CUDNN 8.0.0, tensorrt 7.1.0)

Actually cudnn8.0.0 support was added with this commit: https://github.com/ceccocats/tkDNN/commit/20303ac32ef8e28ca0ff94be633478d7755ae307 You are probably using 8.0.1 Nvidia changes a lot, it is not simple to keep updated. We decided to keep updated with the versions shipped with Jetpack. The current release is: Jetpack 4.4 (CUDA 10.2, CUDNN 8.0.0, tensorrt 7.1.0)
ahmedius2 commented 2020-07-16 03:01:24 +02:00 (Migrated from github.com)

I had the same problem as you and modified the code to be able to compile and run with cuDNN 8.0.0 on jetson nx. Here is the patch: https://drive.google.com/file/d/17f_i7uYClNxyCrL2_QLQXNMDp5OhShKE/view?usp=sharing

I had the same problem as you and modified the code to be able to compile and run with cuDNN 8.0.0 on jetson nx. Here is the patch: https://drive.google.com/file/d/17f_i7uYClNxyCrL2_QLQXNMDp5OhShKE/view?usp=sharing
This repo is archived. You cannot comment on issues.
1 Participants
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: mmr/tkDNN#74