Comments (13)
Hi @MohamadHarastani ,
it seems like nvcc --version
has changed its printed format. We should adapt to it....
Xmipp doesn't need to be compiled in a deactivated environment, indeed it's usually compilated under the scipion environment. However, it could give some problems regarding mixing system libs with environ libs. I should see the error to realize the specific issue.
Regarding the recomemded cuda version. Xmipp alone it's fully tested with cuda-8 and partially tested with cuda<11, for that reason we recomend cuda8 for xmipp (standalone version). However, Scipion has to deal with a lot of packages (xmipp, motioncor, relion, gctf...) and cuda10.1 is compatible with all them. For that reason Scipion recomends cuda10.1.
Thanks Mohamad.
from xmipp.
Hi @dmaluenda
it seems like nvcc --version has changed its printed format. We should adapt to it....
Thanks for the information.
I also find it strange that there is the _bu in the version.
Xmipp doesn't need to be compiled in a deactivated environment, indeed it's usually compilated under the scipion environment. However, it could give some problems regarding mixing system libs with environ libs. I should see the error to realize the specific issue.
Here is the error: I am compiling on Centos7. If I deactivate conda, the installation works.
g++ -o bin/xmipp_classify_analyze_cluster applications/programs/classify_analyze_cluster/classify_analyze_cluster_main.o -L/home/guest/xmipp-bundle/xmipp/src/xmippCore/lib -Llib -Llib -L/home/guest/anaconda3/lib -L/usr/local/cuda -L/usr/local/cuda/lib64 -L/usr/lib -L/usr/lib64/openmpi3/lib -L/usr/lib -L/usr/lib64/openmpi3/lib -L/usr/local/IMOD/lib -lXmipp -lhdf5 -lhdf5_cpp -lfftw3 -lfftw3_threads -ljpeg -ltiff -lfftw3f -lfftw3f_threads -lpthread -lpython3.8 -lXmippCore
lib/libXmipp.so: undefined reference to `non-virtual thunk to H5::H5File::throwException(std::string const&, std::string const&) const'
lib/libXmipp.so: undefined reference to `H5::H5File::throwException(std::string const&, std::string const&) const'
/home/guest/xmipp-bundle/xmipp/src/xmippCore/lib/libXmippCore.so: undefined reference to `H5::H5Location::getFileName() const'
collect2: error: ld returned 1 exit status
scons: *** [bin/xmipp_classify_analyze_cluster] Error 1
scons: building terminated because of errors.
Some error occurred during the compilation of 'xmipp' ('bin/xmipp_classify_analyze_cluster').
Cheers
from xmipp.
Hi @MohamadHarastani
Thanks for reporting this. I'd like to check cuda 11 later this week. I agree with @dmaluenda that they probably changed the format.
We expect sth like Cuda compilation tools, release 8.0, V8.0.61
to be on the last line, where the version we parse is the last group 8.0.61
. See https://github.com/I2PC/xmipp/blob/devel/xmipp#L1412 for more details
As for the build issue you get - looks like the environment does not inherit path to hdf5 for some reason. Check that it's properly set in the xmipp.conf
file (INCDIRFLAGS=..... -I/usr/include/hdf5/serial
). Consider making a backup of the config file, and run ./xmipp config
to regenerate it.
KR
from xmipp.
Hi @DStrelak
Thanks for the details. It's a good idea to test Cuda 11 (many people privilege it, especially Ubuntu 20 users)
I will check the config file tomorrow and let you know.
Regards
from xmipp.
@DStrelak
I tried again to install xmipp using conda (this time with cuda 10.1) and I got the same error.
Here is what INCDIRFLAGS
in xmipp.conf is detecting INCDIRFLAGS=-I../ -I/home/guest/anaconda3/include
I don't know if this file helps ls /home/guest/anaconda3/include >> list.txt
list.txt
from xmipp.
I deavtivated conda, and the installation is okay
Here is what I see in xmipp.conf
INCDIRFLAGS=-I../ -I/usr/include
from xmipp.
Hi @MohamadHarastani ,
my bad, the problem is during linking phase, not the compilation. My best guess that there's some version clash between what you have in includes and what is in the linked library (libhdf5.so) OR the library is found at all (probably the former, as only few bad links are listed).
I think there was another issue involving hdf5. Maybe @pconesa has an idea.
Can you try conda install hdf5 -c defaults
meanwhile?
from xmipp.
Other user has reported problems compiling xmipp on centos related to hdf5. She actually asked for binaries and @MohamadHarastani facilitated them.
I do not know why this was happening and is something to investigte.
from xmipp.
Hi @DStrelak ,
I can try what you suggested in mid-next week. We are forced to stay home and turn off the lab computers due to the hot weather.
I believe that the other user that @pconesa mentioned had a special problem that we can't generalize (installation on a cluster or something). We have Centos7 on several computers in the lab, unfortunately, all have tcsh shell that is not yet compatible with scipion3 conda installation. I changed one of them to bash, scipion3 installs but xmipp doesn't compile on conda (which is why I reported this here).
I will try to do the same with the other computers one after another while we try to solve this issue and let you know what happens.
Cheers
from xmipp.
Hi again,
I have tried this
Can you try
conda install hdf5 -c defaults
meanwhile?
Now I am getting an error that I am not able to workaround (seems that I went in a trap or something have changed).
Here is what I am getting:
./xmipp >> log.txt
log.txt
from xmipp.
Hi @MohamadHarastani ,
it doesn't seem that the hdf5 reinstall helped. Did you keep the log? Did you re-run ./xmipp config
?
From the log, there are two problems - still the hdf5 linking issue and I think that you are on recent Xmipp devel branch. We did some changes in the compilation process on devel branch, which required changes also in the xmippCore. ./xmipp git checkout devel ; ./xmipp git pull
should fix those.
Let me know (via email) if you're available today (till ~17:00 CEST) for a video call.
from xmipp.
Hi @MohamadHarastani ,
it doesn't seem that the hdf5 reinstall helped. Did you keep the log? Did you re-run./xmipp config
?
From the log, there are two problems - still the hdf5 linking issue and I think that you are on recent Xmipp devel branch. We did some changes in the compilation process on devel branch, which required changes also in the xmippCore../xmipp git checkout devel ; ./xmipp git pull
should fix those.Let me know (via email) if you're available today (till ~17:00 CEST) for a video call.
Hi @DStrelak ,
I removed xmipp.conf, pull and checkout.. same issue.
I see that there is a similar issue on Centos7.
I am not in the office today, but I have remote access. I will email you to arrange a video call.
Cheers,
from xmipp.
Cuda 11 support added in #373
from xmipp.
Related Issues (20)
- Not compiling with CUDA 11.4 HOT 1
- Performance of tomo_confidence_map
- Debian package HOT 1
- wrong path to the libcifpp HOT 9
- angular_project_library produces deterministic results when perturbed HOT 2
- Cannot fetch googletest HOT 1
- xmipp.conf's CUDA ignored for compiling cuFFTAdvisor HOT 1
- xmipp_volum_from_pdb gives different contrast according to the pixel size HOT 1
- libcifpp installation error stops xmipp compilation HOT 2
- libcifpp compilation failed HOT 6
- Installer: Popen.poll() returning None in unexpected situations
- non-standard path gcc compile googltest failed in centos7 HOT 1
- New Xmipp installer HOT 3
- python 3.12 error installing HOT 2
- local resolution to B factor has a zero value for residue 1000
- New CMake based compilation script
- error on installation of xmipp through scipion3 HOT 3
- Clean bin without removing xmipp.conf HOT 1
- MADV_HUGEPAGE error HOT 1
- xmipp3-volume consensus fails
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from xmipp.