search

yjxiong / caffe

A fork of Caffe with OpenMPI-based Multi-GPU (mainly data parallel) support for action recognition and more. More documentation please see the original readme.
http://caffe.berkeleyvision.org/
Other
551 stars 153 forks source link

cancelled #217

Closed Helenll closed 5 years ago

Helenll commented 5 years ago

cancelled

yjxiong commented 5 years ago

The docker file for the tsn image is here

https://github.com/yjxiong/temporal-segment-networks/blob/docker/Dockerfile

Its CUDA9, cuDNNv7

yjxiong commented 5 years ago

CUDA is not detected by cmake it is not detected, not unsupported. You should check your environment.

ShawKai666 commented 5 years ago

Thanks for your reply! May I confirm that this repository(https://github.com/yjxiong/caffe) also support CUDA9, cuDNNv7? The docker image is very helpful, but I still need to build caffe in an environment. In ubuntu16.04 I have installed cuda9.0, cudnn7 ,openmpi, and downloaded temporal-segment-networks in advance. So I: $MPI_PREFIX=/usr/local/openmpi bash build_all.sh MPI_ON -- CUDA is not detected by cmake Caffe Configuration Summary -- Dependencies: -- OpenCV : Yes (ver. 2.4.13) -- CUDA : No …... All tools built. Happy experimenting!

I cp the example to Makefile.config, enable USE_CUDNN := 1 and reinstall, but tools/CMakeFiles/caffe.bin.dir/build.make:133: recipe for target 'tools/caffe' failed make[2]: [tools/caffe] Error 1 CMakeFiles/Makefile2:401: recipe for target 'tools/CMakeFiles/caffe.bin.dir/all' failed make[1]: [tools/CMakeFiles/caffe.bin.dir/all] Error 2 ....... Makefile:127: recipe for target 'all' failed make: *** [all] Error 2 Failed to build Caffe. Please check the logs above.

May I exclude the case of not supporting CUDA9, cuDNNv7? Thank you very much for your help!

I have encountered the same problem with you, but my cuda can be detect and ask how you solved it.