Speed Up R Package Installation

Setting up R in a new work environment can take a lot of time because package installation and dependency builds are single-threaded by default.

Install dependencies concurrently

If you are installing multiple source packages, install.packages() can run installs in parallel. The Ncpus argument controls how many concurrent processes are used, so you can keep builds responsive while still saturating your machine during R package installs. [1]

# Use all cores for parallel source installs
install.packages(
  c("data.table", "Rcpp", "sf"),
  Ncpus = parallel::detectCores()
)

# Or set a global default for the session
options(Ncpus = max(1L, parallel::detectCores() - 1L))

Set `MAKEFLAGS` env variable

Many packages compile native code with make. Setting MAKEFLAGS enables parallel compilation, which is often the slowest part of an R package source install. [2]

# Set the number of cores for make-based builds
Sys.setenv(MAKEFLAGS = sprintf("-j%d", parallel::detectCores()))

For a permanent setting, add a line like MAKEFLAGS=-j8 to your ~/.Renviron.

Use CCache

If you build packages with C or C++ extensions regularly, ccache can skip repeated compilation work by caching object files. This is especially effective in CI or when reinstalling the same R packages after small changes. [3]

# Wrap R package compilation (Linux/macOS) for the current session
export CC="ccache gcc"
export CXX="ccache g++"
export CCACHE_DIR="$HOME/.ccache"
R CMD INSTALL ggplot2

# Or use install.packages (still uses CC/CXX if set in the shell)
install.packages("ggplot2")

Other ways to speed things up

Select a faster CRAN mirror close to your location to reduce download timeouts and latency when fetching R packages. [4]

# One-time choice for the current session
chooseCRANmirror()

# Or pin a known fast mirror
options(repos = c(CRAN = "https://cloud.r-project.org"))

On Ubuntu, install prebuilt binaries for R packages when available to avoid compilation overhead. Example for ggplot2: [5]

sudo apt update -qq
sudo apt install --no-install-recommends r-cran-ggplot2

Use Conda/Miniconda with conda-forge packages for a precompiled ecosystem, especially for heavy system dependencies used by R packages. Example for ggplot2: [6]

conda create -n r-fast -c conda-forge r-base r-ggplot2
conda activate r-fast

References

[1] R Core Team. Install packages from repositories or local files. https://stat.ethz.ch/R-manual/R-devel/library/utils/html/install.packages.html

[2] R Core Team. R Installation and Administration. https://cran.r-project.org/doc/manuals/r-release/R-admin.html

[3] Ccache Manual. https://ccache.dev/manual/latest.html

[4] CRAN Mirrors. https://cran.r-project.org/mirrors.html

[5] Ubuntu Packages for R. https://cran.r-project.org/bin/linux/ubuntu/

[6] conda-forge. https://conda-forge.org/