From david.mcenhill at virgin.net Mon Aug 8 08:27:26 2005 From: david.mcenhill at virgin.net (David McEnhill) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Scyld 27BZ-8 and P4 problems Message-ID: <42F779DE.2050907@virgin.net> Hi, first time poster here so I'll try to be concise; I'm currently setting up a small (6 node) cluster for experimenting on. I'm using the "Label Side Up" edition from linux central. The cluster is built up of a number of different processors and machine specs at the moment and it appears to work fine when just using the Athlons in the network, (2x1GHz and an 850MHz) but when i try and wake up the P4's they get to the second stage of booting and then crash with an exeption error. Any help would be greatly appreciated. Regards, Dave. From ghugeha at auburn.edu Sat Aug 20 14:23:25 2005 From: ghugeha at auburn.edu (Harshavardhan Ghuge) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Code execution on compute node Message-ID: Hi, I am am a new user of the scyld beowulf cluster. I am not the root user. I am trying to execute a fortran code in the cluster. The code is not a cluster aware code. The code reads data from a input file and then writes data to an output file. I use the following command to execute the code bpsh 6 ./binary.out where 6 is the node number.I have the binary.out and the input data file stored at two locations. One location is one of the sub-directories on the master node and other is a directory /scratch on the node 6. After the bpsh command the code gets executed but it uses the input data file stored in the sub directories on the master node.It does not use the data file stored on node 6. This happens because the bpsh command uses the binary.out file stored on the master node. The program starts on the master node and then it is passsed to the compute node. What I want is that the code should use the input data file stored in the compute node and also the code should create the output data file in the compute node. I tried to do this by deleting the data files on the master node but then the code does not work. Any suggestions?? From ghugeha at auburn.edu Mon Aug 22 07:51:15 2005 From: ghugeha at auburn.edu (Harshavardhan Ghuge) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Fortran code execution on compute node Message-ID: Hi, I am am a new user of the scyld beowulf cluster. I am trying to execute a fortran code in the cluster. The code is not a cluster aware code. The code reads data from a input file and then writes data to an output file. I use the following command to execute the code bpsh 6 ./binary.out where 6 is the node number.I have the binary.out and the input data file stored at two locations. One location is one of the sub-directories on the master node and other is a directory /scratch on the node 6. After the bpsh command the code gets executed but it uses the input data file stored in the sub directories on the master node.It does not use the data file stored on node 6. This happens because the bpsh command uses the binary.out file stored on the master node. The program starts on the master node and then it is passsed to the compute node. What I want is that the code should use the input data file stored in the compute node and also the code should create the output data file in the compute node. I tried to do this by deleting the data files on the master node but then the code does not work. Any suggestions?? Harshal Ghuge Graduate Research Assistant Aerospace Engineering Auburn University Email ghugeha@auburn.edu From mwill at penguincomputing.com Mon Aug 22 09:11:31 2005 From: mwill at penguincomputing.com (Michael Will) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Fortran code execution on compute node In-Reply-To: References: Message-ID: <4309F933.4030206@penguincomputing.com> Which version of scyld are you using? There used to be a bug that I ran into with Amber, that when executing a file that resided in a different directory than my current working directory, it would open input files from the current directory on the master, but from the executables directory on the compute node. So it behaved differently on the master than on the compute node. I had called it with mpirun ../../bin/foo The workaround until it was fixed was to do a link of the binary into my data directory where I wanted it to execute, so I could call it with mpirun ./foo instead. This was without the complication of having some directories local and some shared by NFS but it might still be the same issue. Michael Will Harshavardhan Ghuge wrote: >Hi, >I am am a new user of the scyld beowulf cluster. I am trying to execute >a fortran code in the cluster. The code is not a cluster aware code. >The code reads data from a input file and then writes data to an output >file. >I use the following command to execute the code >bpsh 6 ./binary.out >where 6 is the node number.I have the binary.out and the input data file >stored at two locations. One location is one of the sub-directories on >the master node and other is a directory /scratch on the node 6. After >the bpsh command the code gets executed but it uses the input data file >stored in the sub directories on the master node.It does not use the >data file stored on node 6. This happens because the bpsh command uses >the binary.out file stored on the master node. The program starts on the >master node and then it is passsed to the compute node. >What I want is that the code should use the input data file stored in >the compute node and also the code should create the output data file in >the compute node. >I tried to do this by deleting the data files on the master node but >then the code does not work. >Any suggestions?? > >Harshal Ghuge >Graduate Research Assistant >Aerospace Engineering >Auburn University >Email ghugeha@auburn.edu >_______________________________________________ >Scyld-users mailing list, Scyld-users@beowulf.org >To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/scyld-users > > -- Michael Will Penguin Computing Corp. Sales Engineer 415-954-2822 415-954-2899 fx mwill@penguincomputing.com From ghugeha at auburn.edu Wed Aug 24 06:36:02 2005 From: ghugeha at auburn.edu (Harshavardhan Ghuge) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Fortran code execution on single compute node Message-ID: Hi, Let me put my problem in a different way. I have a binary stored in a directory named /scratch on a compute node 6. This directory is not on the NFS. I have the input data file required for the binary in the same directory. Is there any command which I can use to execute this binary? (I am asking this because the bpsh command uses the binary stored in the master node) The other way is using the bpsh command. I use the following command- [ghugeha@cicero stat-oplt-unsteady]$ bpsh 6 sh -c "./stat.opltf.v6.out" This binary looks for the input data file in the same directory where it is stored. How can I tell the binary to look for the input data file in a /scratch directory (which is not on the NFS) on compute node 6 ? Can I specify the path of this input data file in my fortran code? Thanks, Harshal From ghugeha at auburn.edu Wed Aug 31 06:48:06 2005 From: ghugeha at auburn.edu (Harshavardhan Ghuge) Date: Tue Nov 9 01:14:28 2010 Subject: [scyld-users] Code execution time Message-ID: Hi, I have executed a Fortran code on scyld beowulf cluster by two ways. In the first method I store the input data file in a subdirectory on the master node. The binary is stored in the same sub directory. Following is the command I use- bpsh 6 sh -c "time ./binary" The program writes data to a output data file which gets stored in the same sub directory on the master node. In this method the input data file is 4 GB. It is used through the NFS.The output data file is not very big. Time required for the code execution is- 10 mins 33.073 sec In the second method I store the input data file in a directory /scratch on compute node 6. This directory is not on the NFS. I want the output data file to be written in the /scratch on compute node 6. So I have specified the path of the input data file and the output data file in the Fortran code itself. I use the following command- bpsh 6 sh -c "time ./binary" Note the binary is still stored on the master node. The input data file stored in the /scratch directory is used and the output data file is created in the /scratch directory. The input data file which is 4 GB is not used though the NFS as it is stored on the node itself. Same is the case for the output data file. The time required for code execution is - 10 min 42.068 sec I expected that the code execution time for the second method will be less than the first method. I thought that in the first method the huge input data file is used through NFS and therefore the code execution time will be more. But the results are other way round. Can somebody tell me why? Harshal