-
On the UCI cluster, the resource request "-pe openmp 64" refers to the number of processors requested. Does that request mean that your commands will use multiple processors?
-
In general, how do you know how many processors, how much RAM, how many files would be required/needed/written by the jobs you are running on the cluster?
-
In order to be a "good citizen", you need to have some idea of much RAM your job requires. In particular, you need to know the "peak" (i.e., maximum) RAM required at any point during execution. Show an example of the shell command that you would use on a Linux machine to measure run time and peak ram usage of an arbitrary command, where the time/peak RAM values are written to a file.
-
What are the units of your answer for number 3?
-
What are the bash commands for the following operations:
- Checking that a file exists
- Checking that a file exists and is not empty
-
How would you use the commands from your answer to 5 to write a work flow for HPC that only runs a job if the expected output file is not present.
-
Would your answer to number 3 work on Apple's OS X operating system? If no, do you have any idea why not?
-
Most of the HPC nodes have 512Gb (gigabytes) of RAM. Let's say you have a job that will require no more than 24Gb of RAM. How would you request resources so that you can run more than one job on a node and not cause nodes to crash? Show an example of a skeleton HPC script as part of your answer. Knowing how to do this is super important and will save you loads of frustration and prevent you from taking out your colleagues' jobs on the cluster, preventing you from getting nasty emails from Harry!!!!!!!!!!!