I have installed ArrayFire 3.8.0 using the binary installer as per the instructions mentioned in the documentation. All dependencies were installed before installing it. NVIDIA CUDA Toolkit 11.3 is also installed. Installation was successful.
To test the installation, I have also build the examples as mentioned in the guide. There was no error during the build process. CPU & CUDA builds are running fine but OpenCL samples are not running. When I execute ./helloworld_opencl example, nothing happens on terminal. No process is shown in top & nvidia-smi.
Please guide me what could be the issue here. Clinfo command is running fine. I am on Ubuntu 20.04.
Thanks
Related
I have been attempting to install the gpuR package as it is a dependency for another package I want to use. However, simply running install.packages('gpuR') generates an error. For context, I am using a very new install of Linux. The error:
.onAttach failed in attachNamespace() for 'gpuR', details:
call: initContexts()
error: ViennaCL: FATAL ERROR: ViennaCL encountered an unknown OpenCL error. Most likely your OpenCL SDK or driver is not installed properly. In some cases, this error is due to an invalid global work size or several kernel compilation errors.
So it says the potential error clearly there. However, the question now lies on how to properly install OpenCL. Using an AskUbuntu post, I ran the following commands:
sudo apt update
sudo apt install ocl-icd-opencl-dev
Still, this gave the same error. I browsed the intel forums a bit, and installation seems to vary. What would be the correct steps to install OpenCL so that gpuR may be used on Ubuntu 18.04?
ocl-icd is only the loader library. You need to install an implementation (that can be loaded by ocl-icd). Where can you get the implementation, depends on what hardware (GPU) you have. I'll assume Intel since you mention their forums. If you have Haswell or older, google for "intel beignet". If you have Broadwell & newer, google "intel compute runtime".
There is an r-base package available for the IBM Power9 CPU architecture (ppc64le) on Ubuntu Linux
and I am wondering if R can really be used on this architecture beyond "does-it-run" experiments.
I did use this search engine to find the r-base package:
https://developer.ibm.com/linuxonpower/open-source-pkgs/
Since this architecture is not officially supported by the R consortium: Does anybody successfully use R on this architecture in a production environment?
Edit 1: RStudio Server seems to be running at least within docker (but there is no official support by RStudio):
A Guide to Enable RStudio on IBM Power System
Running RStudio Server on IBM/Power8
Edit 2: I have found this post at r-devel regarding a solved build failure.
Now a bit of background of my current setup:
I have Python3.3 running on Centos 6. I'm currently working on a web application using Flask that runs on Apache 2.2.15 with mode WSGI 4.5.3 and virtualenv 15.0.2.
pip --version pip 8.1.2 from /usr/local/bin/lib/python3.3/site-packages (python 3.3)
I have installed pysvn with pip and when I run pip show pysvn says
Location: /usr/local/bin/lib/python3.3/site-packages
755 permissions recursively set to /usr/local/bin/lib/python3.3/site-packages. And I passed --system-site-packages argument to virtualenv to use the global site packages.
Even when I try to import the package from python interpreter it does not work. So it is not specific to my virtualenv setup but rather a global problem.
I must mention that other packages installed with pip work perfectly fine (i.e. flask).
I've exhausted all other avenues before coming forward to you guys. Any suggestion would be highly appreciated as I ran out of ideas.
L.E.
I did manage to install it in the end. I'm not completly sure yet why and how but I presume is was compatibility issue.
First of all I have uninstalled svn 1.6+ and installed version 1.8.16 instead which seems to be tested against the latest two versions.
Second, I have uninstalled the troublesome pysvn instance and installed pysvn-1.8.0 workbench "sudo /var/www/FlaskApp/FlaskApp/flask/bin/pip install pysvn-1.8.0.tar.gz". In this case I have installed it my local environment. The 1.9.0 version of pysvn did not work.
L.L.E.
False positive, still doesn't work. I'm going to interact with svn via command line from my script.
L.L.L.E.
After installing svn 1.8.16 and svn-devel along with the rest of dependencies described in the readme file I have managed to successfully install it from the source fallowing the instructions.
Thanks for your help Barry.
pysvn is not available from PyPI because PyPI has no way to allow me to upload pysvn for each supported SVN version. Let alone deal with the issues of installing on a Linux distro given the choices for pysvn dependencies.
(APR, SVN, OpenSSL etc).
Fedora packages pysvn for the Fedora release.
I'm assuming that means it is on RHEL and therefore packaged by CentOS.
(But I do not have RHEL or CentOS to check this on)
If you find that a package is not available for your CentOs is not hard
to build pysvn on a linux distro. Get the source kit and follow the readme.html should get you going.
Barry (pysvn maintainer)
I am trying to install iPDC on a Centos unix laptop.
I am getting a make error when I attempt to install the programme - I have attached a screenshot of my problem.
The command run is sudo make install and I am attempting to install as the root user.
Your installed GTK version is probably too old to support this software. GtkBuilder (a component within GTK) showed up at version 2.12. To find out what version you have, run pkg-config --modversion gtk+-2.0 at the command line. But that version has been around for quite some time. What version Centos are you running? I assume 5, which is quite old.
Upgrading GTK can be tricky, as most of your desktop software relies on it. If you're in for an adventure, the "easiest" would be to upgrade your Centos OS (to 6.x). You might be able to compile a more recent GTK from source and keep it separate from your system GTK, but that will take some patience.
It seems that GTK is not installed.
Try something like: yum install gtk2 or yum install gtk2-devel
I succesfully build Qt 4.8 (without QtWebKit) on CentOS 5.8 (Qt Assistant and the demo's run fine). I also built Qt Creator 2.4.1 against Qt 4.8. I did not observe any errors during the process of building Qt Creator. However, when I attempt to run Qt Creator I receive the following error:
[root#localhost bin]# ./qtcreator
Cannot resolve 'dbus_get_version' in your libdbus-1.
Aborted
Has anybody encountered this error before? How did you fix it? Unfortunately upgrading to a better OS (such as Ubuntu) is not an option for me.
Thanks,
David
QtDBus module loads libdbus-1 library in runtime (via dlopen). That's why you don't have compile/link errors.
You can try either to upgrade dbus package (or whatever package contains libdbus-1.so) or use previous version of Qt.
(In Gentoo QtDBus-4.8 depends on dbus-1.2 and QtDBus-4.7.4 depends on dbus-1.0.2).