My programming experience is about 1 year of C/C++ experience from high school, but I did my research and wrote a simple program with OpenCL a few months ago. I was able to compile and run this on an Apple computer relatively easily with g++ and the --framework option. Now I'm on my Ubuntu machine and I have no idea how to compile it. The correct drivers have been downloaded along with ATI's Stream SDK (I have an ATI Radeon HD5870). Any help would be appreciated!
Try
locate libOpenCL.so
If it is in one of the standard directories (most likely /usr/lib, or /usr/local/lib) you need to replace "--framework OpenCL" with "-lOpenCL". If g++ cannot find the lib you can tell g++ to look in a specific directory by adding "-L/path/to/library".
I wish I had my Linux to be more helpful... It is probably best if you redownload the ati-stream-sdk, after extracting it, open the Terminal and "cd /path/to/extracted/files"; in that directory execute make && sudo make install
make you probably know this from windows, this compiles, whatever needs to be compiled
&& chains commands together, the following commands will only be executed if the first command succeeded
sudo make install this will put the files in the expected places (sudo executes a command with superuser priviledges, you will have to enter your password)
Hope that helps.
You might be missing the dynamic libraries from the dynamic linker configuration.
Search for where the libraries are. Most likely /usr/lib, or /usr/local/lib.
Make sure the path location is also configured at one of these places:
LD_LIBRARY_PATH - you can set it in you environment shell, like .bashrc
/etc/ld.so.conf - you will need to call ldconfig to update the cache and it requires root access to change the file.
Reason
Aside from #bjoernz, my system can't find the libOpenCL.so file
It's because the correct file directory is missing
After searchig over the internet, I found out that libOpenCL.so file can provided by ocl-icd-opencl-dev package
Solution
You just need to install the package mentioned above by typing into cmd
sudo apt update
sudo apt install ocl-icd-opencl-dev
Therefore, libOpenCL.so can be found under /usr/lib/x86_64-linux-gnu/ folder
My System Information
OS: Ubuntu 16.04 LTS
GPU: NVIDIA GeForce GTX 660
GPU Driver: nvidia-375
OpenCL: 1.2
Reference:
[1] How to install libOpenCL.so on ubuntu
[2] How to set up OpenCL in Linux
Related
I am trying to run the metal executable from my zsh terminal in order to meta-analyze GWAS data. I have the executable in the correct directory and have checked that it is not 0MB due to truncation.
Reproducible:
Download the Linux file from http://csg.sph.umich.edu/abecasis/metal/download/
In terminal:
PATH TO EXECUTABLE ./metal
zsh: exec format error: ./metal
You probably install wrong OS of go, for example, you might install go for MacOS in Linux
I solve this problem by installing go for correct OS
The executable has been pre-compiled on a certain distribution with libraries at a certain places at certain versions.
If you have a different distribution, libraries versions, it won't work and you better compile from the source.
Basically what you have to do is to download and extract the sources, go in the folder and execute make. (You will have probably to install make first.)
I think that's not your job at all so maybe you can find a geeky person to help you, because you may stumble upon problems, libraries to install, old versions not supported anymore, new versions not supported yet...
It happened to me when I emptied an executable by mistake.
~>true > a
~>wc -c a
0 a
~>./a
zsh: exec format error: ./a
The setup I have is like this: I have two sets of libraries that are compiled for amd64 (pc) and armelx (ARM). They are both used to cross-compile some software on a build machine.
The first ones (amd64) can be updated without hassle by updating the apt-repository and using apt-get install on the build machine. The packages for ARM however, I don't want to install with apt, because it does not support installing to different directory. If I installed to default directories, the versions could not coexist. Right?
So far, the build machine was updated manually each time there was a new version of the packages, simply by extracting with dpkg -x to a dedicated "fake" footfs directory. This is where the compiler would also look when cross compiling other SW. The problem is, there is no information about these extracted packages or their versions anywhere on the system, right? It should have been in the status file.
My thought was to have these packages installed on this footfs dir with dpkg -i <package.deb> --root=<rootfs>. Would this work? I have a feeling it will not, because the deb packages have no post/pre-remove/install scripts, so it may work for a virgin install somehow, but not for upgrading? Also, what must the rootfs directory structure look like and what must it contain in order for this to work even the first time? Is there a tool to help with this?
Thanks.
Once you have a base armel Debian system, you can actually enter it and run the armel code inside it using something like QEMU. The qemu-arm-static tool (in the qemu-user-static package) can make use of the binfmt_misc capability in Linux to make it so ARM executables are directly run under the QEMU ARM system emulator. So you can run dpkg, apt-get, and so on inside the armel "rootfs" while running on amd64 hardware.
Example:
my_arm_system=/mnt/arm_system
sudo cp /usr/bin/qemu-arm-static "$my_arm_system/usr/bin/"
sudo chroot "$my_arm_system" apt-get update
sudo chroot "$my_arm_system" apt-get install $somepkg
sudo chroot "$my_arm_system" /bin/bash
As for setting up the base armel system in the first place: Debootstrap is the typical method for setting up a Debian base system, whether in a chroot or otherwise. You can use it for installing a base system of a different architecture, but it takes a few extra steps:
distro=jessie # or whatever
echo "Debootstrap phase 1"
sudo mkdir "$my_arm_system"
sudo debootstrap --arch=armel --verbose --foreign "$distro" "$my_arm_system"
sudo cp /usr/bin/qemu-arm-static "$my_arm_system"/usr/bin/
echo "Debootstrap phase 2"
sudo chroot "$my_arm_system" /debootstrap/debootstrap --second-stage
Multistrap is another tool that might be useful; it is intended for setting up Debian environments of one architecture on a host of a different architecture, or for using more complicated APT source combinations. It's not perfect, as it doesn't follow all the deb installation "rules" exactly. It takes some shortcuts/deviations in order to make its job reasonably possible.
I'm trying to setup PHP, apache environment on HP-UX server. While install i'm using usual commands of "./configur, make, make install". Here when I'm trying to install PCRE I got an error like follows.
CDPATH="${ZSH_VERSION+.}:" && cd . && /bin/bash /home/ubuntu/softwares/m4-1.4.17/build-aux/missing aclocal-1.14 -I m4 /home/ubuntu/softwares/m4-1.4.17/build-aux/missing: line 81: aclocal-1.14: command not found WARNING: 'aclocal-1.14' is missing on your system.
You should only need it if you modified 'acinclude.m4' or
'configure.ac' or m4 files included by 'configure.ac'.
The 'aclocal' program is part of the GNU Automake package:
<http://www.gnu.org/software/automake>
It also requires GNU Autoconf, GNU m4 and Perl in order to run:
<http://www.gnu.org/software/autoconf>
<http://www.gnu.org/software/m4/>
<http://www.perl.org/> Makefile:1496: recipe for target 'aclocal.m4' failed make: *** [aclocal.m4] Error 127
So I download latest versions of "m4, autoconf and automake" source and try to install using usual make command.
First I tried to install "automake" it through error asking to install "autoconf"
Then I tried to install autoconf again it ask to install "m4"
Then I tried to install "m4" now it through the same error above listed.
So it became a loop of same set of error not letting me to install.
Can any one help me to sort this issues. Please consider this is a HP-UX unix server so don't recommend the famous ubuntu "apt-get install" command or red hat specific commands.
First read William Pursell's comment to your post (above). If you still need to install the autotools ...
Check to see what, if any, autotools you may already have installed by typing: m4 --versionand autoconf --versionand automake --version.
You should use HP-UX's package manager. It's called Software Distributor (SD). http://en.wikipedia.org/wiki/Software_Distributor
HP-UX's FAQ 5.9 explains how to handle dependencies using depothelper. http://hpux.connect.org.uk/hppd/answers/5-9.html
Here is where you find the correct autotool packages (autoconf, automake, libtool) for HP-UX. Install these HP-UX packages using HP-UX's native package manager instead of compiling from source. http://hpux.connect.org.uk/hppd/packages.html
I was facing the same problem with m4. In my case, the problem was I was transferring all the source files via scp to a server.
When I tried to configure, make and make install through ssh, this kept happening. I believe something did not transfer the way it was supposed to.
The problem was solved by manually transferring the files
through a USB.
It's not a perfect solution (it implies physical access to the server) but it works.
I have sample "Hello, World!" code from the net and I want to run it on the GPU on my university's server. When I type "gcc main.c," it responds with:
CL/cl.h: No such file or directory
What should I do? How can I have this header file?
Are you using Ubuntu or Debian distro? Then you could use this package to solve the problem with missing header file:
apt-get install opencl-headers
You must install opencl library to solve linking issues using that Debian and Ubuntu package:
apt-get install ocl-icd-libopencl1
You can also use these nonfree libraries: nvidia-libopencl1 (Debian) or nvidia-libopencl1-xx (Ubuntu).
Make sure you have the appropriate toolkit installed.
This depends on what you intend running your code on. If you have an NVidia card then you need to download and install the CUDA-toolkit which also contains the necessary binaries and libraries for opencl.
Are you running Linux? If you believe you already have OpenCL installed it could be that it is found at a different location than the standard /usr/include. Type the following and see what results you get:
find / -iname cl.h 2>/dev/null
On my laptop for example, the header is found at /usr/local/cuda-5.5/include. If its the case were your header file is at a different location you simply have to specify the path during complication
g++ -I/usr/local/cuda-5.5/include main.c -lOpenCL
Alternatively, you can create a symbolic link from the path to /usr/include:
ln -s /usr/local/cuda-5.5/include/CL /usr/include
I am trying to start working with OpenCL. I have two NVidia graphics card, I installed "developer driver" as well as SDK from NVidia website. I compiled the demos but when I run
./oclDeviceQuery
I see:
OpenCL SW Info:
Error -1001 in clGetPlatformIDs Call
!!!
How can I fix it? Does it mean my nvidia cards cannot be detected? I am running Ubuntu 10.10 and X server works properly with nvidia driver.
I am pretty sure the problem is not related to file permissions as it doesn't work with sudo either.
In my case I have solved it by installing nvidia-modprobe package available in ubuntu (utopic/multiverse). And the driver itself (v346) was installed from https://launchpad.net/~mamarley/+archive/ubuntu/nvidia
Concretely, I have installed nvidia-opencl-icd-346, nvidia-libopencl1-346, nvidia-346-uvm, nvidia-346 and libcuda1-346. Not sure if they are all needed for OpenCL.
This is a result of not installing the ICD portion of Nvidia's openCL runtime. The ICD profile will instruct your application of the different openCL implementations installed on the system as multiple implementations from different vendors can coexist. Whe your application does not find the ICD information it gives the Error -1001.
Run your program as root. In case of success: you have trouble with cl_khr_icd- extension to load the vendor driver.
If you not running X11, you have to create device files manually or by (boot-)script:
ERROR: clGetPlatformIDs -1001 when running OpenCL code (Linux)
Same problem for me on a Linux system. Solution is to add the user to the video group:
# sudo usermod -aG video your-user-name
Since I just spend a couple of hours on this, I thought I would share:
I got the error because I was connected to the machine per remote desktop (mstsc). On the machine itself everything worked fine.
I have been told that it should work with TeamViewer by the way.
Dont know if you ever solved this problem, but I had the same issue and solved it in this post: ERROR: clGetPlatformIDs -1001 when running OpenCL code (Linux)
Hope it helps!
I have solved it in Ubuntu 13.10 saucy for intel opencl by created link:
sudo ln -s /opt/intel/opencl-1.2-3.2.1.16712/etc/intel64.icd /etc/OpenCL/vendors/nvidia.icd
I just ran into this problem on ubuntu 14.04 and I could not find ANY working answers anywhere online including this thread (though this was the first to show up on google). What ended up working for me was to remove ALL previous nvidia software and then to reinstall it using the .run file provided on the nvidia website. Installing the components through apt-get seems to fail for some reason.
1) Download CUDA .run file: https://developer.nvidia.com/cuda-downloads
2) Purge all previous nvidia packages
sudo apt-get purge nvidia-*
3) Install all run file components (you will likely have to stop X or restart in recovery mode to run this)
sudo sh cuda_X.X.XX_linux.run
This is because OpenCL has the same brain damaged one library per vendor setup that OpenGL has. A likely reason for the -1001 error is that you have compiled with a different library than the linker is trying to dynamically load.
So see if this is the problem run:
$ ldd oclDeviceQuery
...
libOpenCL.so.1 => important path here (0x00007fe2c17fb000)
...
Does the path point towards the NVidia-provided libOpenCL.so.1 file? If it doesn't, you should recompile the program with an -L parameter pointing towards the directory containing NVidia's libOpenCL.so.1. If you can't do that, you can override the linker's path like this:
$ LD_LIBRARY_PATH=/path/to/nvidias/lib ./oclDeviceQuery
For me, I was missing the CUDA OpenCL library, Running sudo apt install cuda-opencl-dev-12-0 solved it.
You should get number of platforms, allocate the memory for platforms, again get this platforms and then create context from this platform. There is good example:
http://developer.amd.com/support/KnowledgeBase/Lists/KnowledgeBase/DispForm.aspx?ID=71
This might be due to querying clGetPlatformIDs by multiple threads at the same time