| 1 |
<!DOCTYPE html> |
| 2 |
<html lang="en"> |
| 3 |
<head> |
| 4 |
<meta charset="UTF-8"> |
| 5 |
<title>Parallel Workers on High-Performance Compute Environments</title> |
| 6 |
<style> |
| 7 |
body { |
| 8 |
font-family: sans-serif; |
| 9 |
line-height: 1.6; |
| 10 |
padding-left: 3ex; |
| 11 |
padding-right: 3ex; |
| 12 |
background-color: white; |
| 13 |
color: black; |
| 14 |
} |
| 15 |
|
| 16 |
a { |
| 17 |
color: #4183C4; |
| 18 |
text-decoration: none; |
| 19 |
} |
| 20 |
|
| 21 |
h1, h2, h3 { |
| 22 |
margin: 2ex 0 1ex; |
| 23 |
padding: 0; |
| 24 |
font-weight: bold; |
| 25 |
-webkit-font-smoothing: antialiased; |
| 26 |
cursor: text; |
| 27 |
position: relative; |
| 28 |
} |
| 29 |
|
| 30 |
h2 { |
| 31 |
border-bottom: 1px solid #cccccc; |
| 32 |
} |
| 33 |
|
| 34 |
code { |
| 35 |
margin: 0 2px; |
| 36 |
padding: 0 5px; |
| 37 |
white-space: nowrap; |
| 38 |
border: 1px solid #eaeaea; |
| 39 |
background-color: #f8f8f8; |
| 40 |
border-radius: 3px; |
| 41 |
} |
| 42 |
|
| 43 |
pre code { |
| 44 |
margin: 0; |
| 45 |
padding: 0; |
| 46 |
white-space: pre; |
| 47 |
border: none; |
| 48 |
background: transparent; |
| 49 |
} |
| 50 |
|
| 51 |
pre { |
| 52 |
background-color: #f8f8f8; |
| 53 |
border: 1px solid #cccccc; |
| 54 |
line-height: 2.5x; |
| 55 |
overflow: auto; |
| 56 |
padding: 0.6ex 1ex; |
| 57 |
border-radius: 3px; |
| 58 |
} |
| 59 |
|
| 60 |
pre code { |
| 61 |
background-color: transparent; |
| 62 |
border: none; |
| 63 |
} |
| 64 |
</style> |
| 65 |
</head> |
| 66 |
<body> |
| 67 |
<h1>Parallel Workers on High-Performance Compute Environments</h1> |
| 68 |
<!-- |
| 69 |
%\VignetteIndexEntry{Parallel Workers on High-Performance Compute Environments} |
| 70 |
%\VignetteAuthor{Henrik Bengtsson} |
| 71 |
%\VignetteKeyword{R} |
| 72 |
%\VignetteKeyword{package} |
| 73 |
%\VignetteKeyword{vignette} |
| 74 |
%\VignetteKeyword{HPC} |
| 75 |
%\VignetteEngine{parallelly::selfonly} |
| 76 |
--> |
| 77 |
<h1>Introduction</h1> |
| 78 |
<p>This vignette illustrates how to launch parallel workers via job |
| 79 |
schedulers running in high-performance compute (HPC) environments.</p> |
| 80 |
<h1>Examples</h1> |
| 81 |
<h2>Example: Launch parallel workers via the Grid Engine job scheduler</h2> |
| 82 |
<p>‘Grid Engine’ is a high-performance compute (HPC) job scheduler where |
| 83 |
one can request compute resources on multiple nodes, each running |
| 84 |
multiple cores. Examples of Grid Engine schedulers are Oracle Grid |
| 85 |
Engine (formerly Sun Grid Engine), Univa Grid Engine, and Son of Grid |
| 86 |
Engine - all commonly referred to as SGE schedulers. Each SGE cluster |
| 87 |
may have its own configuration with its own way of requesting |
| 88 |
parallel slots.</p> |
| 89 |
<p>Consider the following two files: <code>script.sh</code> and <code>script.R</code>.</p> |
| 90 |
<p>script.sh:</p> |
| 91 |
<pre><code class="language-sh">#! /usr/bin/env bash |
| 92 |
#$ -cwd ## Run in current working directory |
| 93 |
#$ -j y ## Merge stdout and stderr |
| 94 |
#$ -l mem_free=100M ## 100 MiB RAM per slot |
| 95 |
#$ -l h_rt=00:10:00 ## 10 minutes runtime |
| 96 |
#$ -pe mpi 8 ## 8 compute slots |
| 97 |
|
| 98 |
echo "Information on R:" |
| 99 |
Rscript --version |
| 100 |
|
| 101 |
echo "Running R script:" |
| 102 |
Rscript script.R |
| 103 |
</code></pre> |
| 104 |
<p>script.R:</p> |
| 105 |
<pre><code class="language-r">library(parallelly) |
| 106 |
library(parallel) |
| 107 |
|
| 108 |
cl <- makeClusterPSOCK( |
| 109 |
availableWorkers(), |
| 110 |
rshcmd = "qrsh", rshopts = c("-inherit", "-nostdin", "-V") |
| 111 |
) |
| 112 |
print(cl) |
| 113 |
|
| 114 |
# Perform calculations in parallel |
| 115 |
X <- 1:100 |
| 116 |
y <- parLapply(cl = cl, X, fun = sqrt) |
| 117 |
y <- unlist(y) |
| 118 |
z <- sum(y) |
| 119 |
print(z) |
| 120 |
|
| 121 |
stopCluster(cl) |
| 122 |
</code></pre> |
| 123 |
<p>The <code>script.sh</code> file is a job script that we submit to the scheduler |
| 124 |
that runs the R script <code>script.R</code> when launched. If we submit |
| 125 |
<code>script.sh</code> as:</p> |
| 126 |
<pre><code class="language-sh">$ qsub script.sh |
| 127 |
</code></pre> |
| 128 |
<p>it will by default request eight slots - on one or more machines, |
| 129 |
which then R and <strong>parallelly</strong> will set up a parallel cluster |
| 130 |
on. Exactly on which machines depends on where the job scheduler finds |
| 131 |
these requested slots.</p> |
| 132 |
<p>Here is the output from one such run, where the scheduler happened to |
| 133 |
allot the slots across three machines:</p> |
| 134 |
<pre><code class="language-sh">Information on R: |
| 135 |
Rscript (R) version 4.5.2 (2025-10-31) |
| 136 |
Running R script: |
| 137 |
Socket cluster with 8 nodes where 4 nodes are on host ‘localhost’ |
| 138 |
(R version 4.5.2 (2025-10-31), platform x86_64-pc-linux-gnu), 3 |
| 139 |
nodes are on host ‘qb3-id130’ (R version 4.5.2 (2025-10-31), |
| 140 |
platform x86_64-pc-linux-gnu), 1 node is on host ‘qb3-as16’ (R |
| 141 |
version 4.5.2 (2025-10-31), platform x86_64-pc-linux-gnu) |
| 142 |
[1] 671.4629 |
| 143 |
</code></pre> |
| 144 |
<h2>Example: Launch parallel workers via the Fujitsu Technical Computing Suite job scheduler</h2> |
| 145 |
<p>The ‘Fujitsu Technical Computing Suite’ is a high-performance compute |
| 146 |
(HPC) job scheduler where one can request compute resources on |
| 147 |
multiple nodes, each running multiple cores.</p> |
| 148 |
<p>Consider the following two files: <code>script.sh</code> and <code>script.R</code>.</p> |
| 149 |
<p>script.sh:</p> |
| 150 |
<pre><code class="language-sh">#! /usr/bin/env bash |
| 151 |
|
| 152 |
echo "Information on R:" |
| 153 |
Rscript --version |
| 154 |
|
| 155 |
echo "Running R script:" |
| 156 |
Rscript script.R |
| 157 |
</code></pre> |
| 158 |
<p>script.R:</p> |
| 159 |
<pre><code class="language-r">library(parallelly) |
| 160 |
library(parallel) |
| 161 |
|
| 162 |
cl <- makeClusterPSOCK( |
| 163 |
availableWorkers(), |
| 164 |
rshcmd = "pjrsh" |
| 165 |
) |
| 166 |
print(cl) |
| 167 |
|
| 168 |
# Perform calculations in parallel |
| 169 |
X <- 1:100 |
| 170 |
y <- parLapply(cl = cl, X, fun = sqrt) |
| 171 |
y <- unlist(y) |
| 172 |
z <- sum(y) |
| 173 |
print(z) |
| 174 |
|
| 175 |
stopCluster(cl) |
| 176 |
</code></pre> |
| 177 |
<p>The <code>script.sh</code> file is a job script that we submit to the scheduler |
| 178 |
that runs the R script <code>script.R</code> when launched. We can submit |
| 179 |
<code>script.sh</code> as:</p> |
| 180 |
<pre><code class="language-sh">$ pjsub -L vnode=3 -L vnode-core=18 script.sh |
| 181 |
</code></pre> |
| 182 |
<p>to request 18 CPU cores on three compute nodes, which in total |
| 183 |
requests 3*18=54 compute slots.</p> |
| 184 |
</body> |
| 185 |
</html> |