What steps will reproduce the problem?
1. Run one of the samples using the provided sh script.
2.
3.
What is the expected output? What do you see instead?
Sep 04, 2014 3:57:57 PM com.amd.aparapi.OpenCLJNI <clinit>
FEIN: arch = amd64
Sep 04, 2014 3:57:57 PM com.amd.aparapi.OpenCLJNI <clinit>
FEIN: attempting to load aparapi shared lib aparapi_x86_64
Check your environment. Failed to load aparapi native library aparapi_x86_64 or
possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure
that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).
Machine contains 0 OpenCL platforms
OpenCLDevice.best() returned null!
OpenCLDevice.firstCPU() returned null!
OpenCLDevice.firstGPU() returned null!
What version of the product are you using? On what operating system?
Aparapi_2013_01_23_linux_x86_64 binary dist.
OS is CentOS 5.10 with CUDA 6.0 (4x Geforce GTX480)
OpenJDK 64-Bit Server VM
Please provide any additional information below.
-Djava.library.path is set to aparapi folder (where libaparapi_x86_64.so is
located)
LD_LIBRARY_PATH contains /usr/local/cuda/lib64 (where libOpenCL.so is located)
Usually, I ssh into the machine (it is a node of our cluster), but I
encountered the same behavior while using VNC. I have also tested it on another
GPU cluster that uses the SLURM scheduler - with the same behavior.
Original issue reported on code.google.com by SteffenE...@gmail.com on 4 Sep 2014 at 2:20
Original issue reported on code.google.com by
SteffenE...@gmail.comon 4 Sep 2014 at 2:20