incompatible with cudnn 4?

[JieSong89]

@tpfister

Hi,
I try to compile with cudnn 4 but seems incompatible.
Have you tried with cudnn 4?

[tpfister]

Hi,

This was forked off Caffe September 2015 so it doesn't contain cudnn4 support.

I'd be happy to accept pull requests for cudnn4 support / merges with latest Caffe master if you need it! :)

Tomas

[harshal2802]

With which cuda version and toolkit version will it work perfectly?
as i am currently having Cuda compilation tools, release 7.5, V7.5.17 installed in my PC.
kindly help me with configuration..

[joefang66]

cudnn v2 works for this project.

I have run the demo with Matlab, but when try to train the network with command "sh train_heatmap.sh heatmap-flic-fusion 1", the following error is reported:

terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injectorboost::bad_lexical_cast >'
what(): bad lexical cast: source type value could not be interpreted as target
*** Aborted at 1489413106 (unix time) try "date -d @1489413106" if you are using GNU date ***
PC: @ 0x7ffb16222c37 (unknown)
*** SIGABRT (@0x3e800001bda) received by PID 7130 (TID 0x7ffb18284a40) from PID 7130; stack trace: ***
@ 0x7ffb16222cb0 (unknown)
@ 0x7ffb16222c37 (unknown)
@ 0x7ffb16226028 (unknown)
@ 0x7ffb17e1d535 (unknown)
@ 0x7ffb17e1b6d6 (unknown)
@ 0x7ffb17e1b703 (unknown)
@ 0x7ffb17e1b922 (unknown)
@ 0x40c331 boost::throw_exception<>()
@ 0x4078e8 get_gpus()
@ 0x408cc5 train()
@ 0x406bd1 main
@ 0x7ffb1620df45 (unknown)
@ 0x40736b (unknown)
@ 0x0 (unknown)
Aborted (core dumped)

Could you please check the error? Thanks.

[xizero00]

hi all:
@tpfister @JieSong89 @harshal2802 @pluskid
I have update caffe with the latest version and create a pull request.
@tpfister Would you please merge it.
The fork in my repo is here
https://github.com/djangogo/caffe-heatmap