NVidia hardware is GPU-based math hardware, the chips in high-end 3d graphics cards. Cuda is the name of the NVidia development environment. Your Hadoop job want to run code on an NVidia card.
I'm guessing 'jcuda' is a utility program for the CUDA tools. What is your code written in? We need more information on what program you are running, where it came from, what programs it expects, etc. Lance On Feb 28, 2011, at 1:32 PM, Greg Roelofs wrote: > Adarsh Sharma <[email protected]> wrote: > >> But Still don't know why it fails in Map-reduce job. > >> [hadoop@ws37-mah-lin hadoop-0.20.2]$ bin/hadoop jar wordcount1.jar >> org.myorg.WordCount /user/hadoop/gutenberg /user/hadoop/output1 >> 11/02/28 15:01:45 INFO input.FileInputFormat: Total input paths to >> process : 3 >> 11/02/28 15:01:45 INFO mapred.JobClient: Running job: job_201102281104_0006 >> 11/02/28 15:01:46 INFO mapred.JobClient: map 0% reduce 0% >> 11/02/28 15:01:56 INFO mapred.JobClient: Task Id : >> attempt_201102281104_0006_m_000000_0, Status : FAILED >> java.lang.RuntimeException: java.lang.reflect.InvocationTargetException > [...] >> Caused by: java.lang.UnsatisfiedLinkError: no jcuda in java.library.path > > That would be why. I don't know if "jcuda" is an external command, a > shared library, or a jar of some sort, but Hadoop wants it in the Java > library path (probably on all nodes). > > Greg
