Running R on Cypress

R Modules

You can list the versions of R available on Cypress as modules with the module avail command as in the following example.

[tulaneID@cypress1 ~]$module avail R
--------------------------------------- /share/apps/modulefiles ----------------------------------------
R/3.1.2(default) R/3.2.5-intel    R/3.4.1-intel    R/3.6.1-intel
R/3.2.4          R/3.3.1-intel    R/3.5.2-intel    ROOT/5.34.36

Observe in the output above that some R modules have names ending with the string 'intel'. These modules have been constructed with links to Intel's Math Kernel Library (MKL) for performing certain computations using the Xeon Phi coprocessors. See cypress/XeonPhi.

Running R Interactively

For Workshop

If you use a temporary workshop account, do this.

export MY_PARTITION=workshop
export MY_QUEUE=workshop

Start an interactive session using idev

[tulaneID@cypress1 ~]$ idev 
Requesting 1 node(s)  task(s) to workshop queue of workshop partition
1 task(s)/node, 20 cpu(s)/task, 2 MIC device(s)/node
Time: 0 (hr) 60 (min).
Submitted batch job 1164332
JOBID=1164332 begin on cypress01-121
--> Creating interactive terminal session (login) on node cypress01-121.
--> You have 0 (hr) 60 (min).
--> Assigned Host List : /tmp/idev_nodes_file_tuhpc002
Last login: Wed Aug 21 15:56:37 2019 from cypress1.cm.cluster
[tulaneID@cypress01-121 ~]$ 

For Workshop

If you use a temporary workshop account, in order to use only 2 cpu's per node and thus allow for sharing the few available nodes in the interactive partition among many users, do this.

[tulaneID@cypress1 ~]$ idev -c 2

Load the R module

[tulaneID@cypress01-121 ~]$ module load R/3.2.4
[tulaneID@cypress01-121 ~]$ module list
Currently Loaded Modulefiles:
  1) slurm/14.03.0       3) bbcp/amd64_rhel60
  2) idev                4) R/3.2.4

Run R in the command line window

[tulaneID@cypress01-121 ~]$R

R version 3.2.4 (2016-03-10) -- "Very Secure Dishes"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

  Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

>

R Package Dependency

In order to run any of the SLURM scripts below, we would submit the SLURM script via sbatch command, but in general we would not expect the code to run successfully the very first time with out additional setup. We can try running the script without any additional setup, but we can expect to get the error such as in the following R session.

> library(doParallel)
Error in library(doParallel) : there is no package called ‘doParallel’

Thus we should first ensure that the required R package, in this case the R package doParallel, is available and installed in your environment. For a range of options for installing R packages - depending on the desired level of reproducibility, see the section Installing R Packages on Cypress.

For Workshop : If you use a temporary workshop account, use Alternative 1 - responding to the R prompts as needed - for installing R packages such as in the following.

Thus we need to install the R package doParallel.

> install.packages("doParallel")
...
(respond to prompts as needed)
...
> library(doParallel)
Loading required package: foreach
Loading required package: iterators
Loading required package: parallel
> q()
Save workspace image? [y/n/c]: n
[tuhpc002@cypress01-121 R]$ exit

Thus, now that we have resolved our package dependency, we can expect future jobs requiring doParallel to run without errors.

Running a R script in Batch mode

Besides running R interactively in an idev session, you can also submit your R job to the batch nodes (compute nodes) on Cypress. Inside your SLURM script, include a command to load the desired R module. Then invoke the Rscript command on your R script.

#!/bin/bash
#SBATCH --qos=normal            # Quality of Service
#SBATCH --job-name=R            # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1     # Number of tasks (MPI processes)
#SBATCH --cpus-per-task=1       # Number of threads per task (OMP threads)

module load R/3.2.4
Rscript myRscript.R

For Workshop : If you use a temporary workshop account, modify the SLURM script like:

#!/bin/bash
#SBATCH --partition=workshop    # Partition
#SBATCH --qos=workshop          # Quality of Service
##SBATCH --qos=normal          ### Quality of Service (like a queue in PBS)
#SBATCH --job-name=R            # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1     # Number of tasks (MPI processes)
#SBATCH --cpus-per-task=1       # Number of threads per task (OMP threads)

module load R/3.2.4
Rscript myRscript.R

Running a Parallel R Job

Starting with version 2.14.0, R has offered direct support for parallel computation through the "parallel" package. We will present two examples of running a parallel job of BATCH mode. They differ in the ways in which they communicate the number of cores reserved by SLURM to R. Both are based on code found in ​"Getting Started with doParallel and foreach" by Steve Weston and Rich Calaway and modified by ​The University of Chicago Resource Computing Center.

Passing (SLURM) Environment Variables

In the first example, we will use the built in R function Sys.getenv( ) to get the SLURM environment variable from the operating system.

Edit the new file bootstrap.R to contain the following code.

#Based on code from the UCRCC website

library(doParallel)

# use the environment variable SLURM_CPUS_PER_TASK to set the number of cores
registerDoParallel(cores=(Sys.getenv("SLURM_CPUS_PER_TASK")))

# Bootstrapping iteration example
x <- iris[which(iris[,5] != "setosa"), c(1,5)]
iterations <- 10000# Number of iterations to run

# Parallel version of code 
# Note the '%dopar%' instruction
part <- system.time({
  r <- foreach(icount(iterations), .combine=cbind) %dopar% {
    ind <- sample(100, 100, replace=TRUE)
    result1 <- glm(x[ind,2]~x[ind,1], family=binomial(logit))
    coefficients(result1)
  }
})[3]

# Shows the number of Parallel Workers to be used
getDoParWorkers()
# Executes the functions
part

The above script will obtain the number of tasks per node via the SLURM environment variable SLURM_CPUS_PER_TASK set in our SLURM script and will pass that value to the registerDoParallel( ) function. To implement this we need only set the correct parameters in our SLURM script. Suppose we wanted to use 16 cores. Then the correct SLURM script would be as follows.

#!/bin/bash
#SBATCH --qos=normal            # Quality of Service
#SBATCH --job-name=R            # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1    # Number of Tasks per Node
#SBATCH --cpus-per-task=16      # Number of threads per task (OMP threads)

module load R/3.2.4

Rscript bootstrap.R

For Workshop : If you use a temporary workshop account, modify the SLURM script like:

#!/bin/bash
#SBATCH --partition=workshop    # Partition
#SBATCH --qos=workshop          # Quality of Service
##SBATCH --qos=normal          ### Quality of Service (like a queue in PBS)
#SBATCH --job-name=R            # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1    # Number of Tasks per Node
#SBATCH --cpus-per-task=16      # Number of threads per task (OMP threads)

module load R/3.2.4

Rscript bootstrap.R

Edit the new file bootstrap.sh to contain the above SLURM script code and submit as shown below.

Also, note that since we did not specify an output file in the SLURM script, the output will be written to slurm-<JobNumber>.out. For example:

[tulaneID@cypress2 R]$ sbatch bootstrap.sh
Submitted batch job 774081
[tulaneID@cypress2 R]$ cat slurm-774081.out
Loading required package: foreach
Loading required package: iterators
Loading required package: parallel
[1] "16"
elapsed
  2.954
[tulaneID@cypress2 R]$

Passing Parameters

The disadvantage of the above approach is that it is system specific. If we move our code to a machine that uses PBS-Torque as it's manager (sphynx for example) we have to change our source code. An alternative method that results in a more portable code base uses command line arguments to pass the value of our environment variables into the script.

Edit the new file bootstrapWargs.R to contain the following code.

#Based on code from the UCRCC website

library(doParallel)
# Enable command line arguments
args<-commandArgs(TRUE)

# use the first command line argument to set the number of cores
registerDoParallel(cores=(as.integer(args[1])))

# Bootstrapping iteration example
x <- iris[which(iris[,5] != "setosa"), c(1,5)]
iterations <- 10000# Number of iterations to run

# Parallel version of code 
# Note the '%dopar%' instruction
part <- system.time({
  r <- foreach(icount(iterations), .combine=cbind) %dopar% {
    ind <- sample(100, 100, replace=TRUE)
    result1 <- glm(x[ind,2]~x[ind,1], family=binomial(logit))
    coefficients(result1)
  }
})[3]

# Shows the number of Parallel Workers to be used
getDoParWorkers()
# Executes the functions
part

Note the use of args←commandArgs(TRUE) and of as.integer(args[1]). This allows us to pass in a value from the command line when we call the script and the number of cores will be set to that value. Using the same basic submission script as last time, we need only pass the value of the correct SLRUM environment variable to the script at runtime.

#!/bin/bash
#SBATCH --qos=normal            # Quality of Service
#SBATCH --job-name=R       # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1    # Number of Tasks per Node
#SBATCH --cpus-per-task=16      # Number of threads per task (OMP threads)

module load R/3.2.4

Rscript bootstrapWargs.R $SLURM_CPUS_PER_TASK 

For Workshop : If you use a temporary workshop account, modify the SLURM script like:

#!/bin/bash
#SBATCH --partition=workshop    # Partition
#SBATCH --qos=workshop          # Quality of Service
##SBATCH --qos=normal          ### Quality of Service (like a queue in PBS)
#SBATCH --job-name=R       # Job Name
#SBATCH --time=00:01:00         # WallTime
#SBATCH --nodes=1               # Number of Nodes
#SBATCH --ntasks-per-node=1    # Number of Tasks per Node
#SBATCH --cpus-per-task=16      # Number of threads per task (OMP threads)

module load R/3.2.4

Rscript bootstrapWargs.R $SLURM_CPUS_PER_TASK 

Edit the new file bootstrapWargs.sh to contain the above SLURM script code.

Now submit as in the following.

[tulaneID@cypress1 ~]$ sbatch bootstrapWargs.sh 
Submitted batch job 52481
[tulaneID@cypress1 ~]$ cat slurm-52481.out 
Loading required package: foreach
Loading required package: iterators
Loading required package: parallel
[1] "16"
elapsed 
  3.282 
[tulaneID@cypress1 ~]$ 

Installing R Packages on Cypress

See here.

Next Section: Running Python on Cypress