NCSA project utilizing the parallel programming skills to improve performance.
- Receive the formula of CowVal formula and calculate the value
- Compare the result with the Matlab Code
- Read the paper and get the ME 305 formula
- Apply concurrent programming on the model
- Improve the running efficiency of the model
- Log into HTC
- Create ME 305 function
- Create a new column cumulate the value of the milk production
- Change the sorting order of the herd(line:490) [9->10]
- Rank the Cow by the value of herd (The value of NPV go up by approximately 300-400)
- Try to run it on HTC
- Finish the first version of the sub and create simplified submission
- Future and more work, integrate Beta into the script file and make it easier to use
- Fix the bug of Exponential Function
- Introduce Beta
- General procedure to run a job on HTC
- Move to the src filecd
cd NCSA_Parllel_herd/JAVAPROJECT/src/
- Compile the java file with javac
javac *.java
- submit a job
condor_submit herd.cmd
- Move to the src filecd
- Useful commands
condor_q
[check the current status of the job]condor_rm [9239]
(campus cluster number) [remove the job by number]condor_rm [9239] | grep [9239]
[When there are a lot of jobs and find to find a specfic one]
####################
#
# Running a mutiple threads version of herd simulation by Professor Rebecca Smith
# Execute the main class named herdRun
#
####################
universe = java
executable = herdRun.class
transfer_input_files = herdHelperFunc$1.class,herdHelperFunc.class,threadHerd.class,herdParameter.class,herd_13shedExcel.csv
arguments = herdRun
output = herdRun.output
error = herdRun.error
Log = myexe.log
should_transfer_files = YES
when_to_transfer_output = ON_EXIT
notification = always
queue
- Error file is stored in herdRun.error and output is stored in herdRun.output(open with cat)
- Transfer_input_files is the sub-java classes that need to be specified for HTC to run
- Executable is the java classes with main
- Notification can be set to <Always|Error|Complete>