How to Use Multiple Cores to Make Gganimate Faster

How to use multiple cores to make gganimate faster

This is a pull request, meaning that the code is available on GitHub as a branch, but hasn't yet been merged in gganimate master.

You could clone it or copy the modified package directory on your system.

Then :

• make sure that devtools package is installed
• open gganimate.Rproj
• run devtools::load_all(".")

The parallel version is ready to run :

anim <- ggplot(mtcars, aes(mpg, disp)) +
  transition_states(gear, transition_length = 2, state_length = 1) +
  enter_fade() +
  exit_fade()

future::plan("sequential")  ## default
t0 <- system.time(animate(anim))
print(t0)

#   user        system      total 
#   4.384615    1.360656    1.893855 

future::plan("multiprocess", workers = 4L)
t1 <- system.time(animate(anim))
#   user        system      total 
#   1.30        0.61        3.58 

print(t0 / t1)
#   user        system      total 
#   4.384615    1.360656    1.893855 

To avoid load_all you could open the DESCRIPTION file and rename the package :

Package: gganimateparallel
Type: Package
Title: A Grammar of Animated Graphics
...

As vignettes seem to have difficulties to build, you can just remove the vignettes directory.

Then RStudio / Build / Install and restart or from the package's directory

Rcmd.exe INSTALL --no-multiarch --with-keep.source .

gganimateparallel is now available on your system as a library.

Credits @HenrikBengtsson for the incredible job done on future!

gganimate parallel - Provided file does not exist error

From trial and error, the following seems to be the problem: the package needs to be build from the GitHub pull request (a) (as described here) and needs to be named gganimate to satisfy internal dependencies.

After setting up the environment correctly, parallel processing seems to work without using the plan(multisession, workers=7) command, but using it appears to throw an error.

(see the discussion under the question for more details)

How to manage parallel processing with animated ggplot2-plot?

So, my solution:

• split dates in ncores periods

• get the plot for each period and save it as a GIF

• read back all GIF and combine them

---

########################################################################
# setup
########################################################################

# creating some sample data for one year
# 4 categories; each category has a specific value per day
set.seed(1)
x <- data.frame(
  rep(as.Date((Sys.Date()-364):Sys.Date(), origin="1970-01-01"),4),
  c(rep("cat01",length.out=365),
    rep("cat02",length.out=365),
    rep("cat03",length.out=365),
    rep("cat04",length.out=365)),
  sample(0:50,365*4, replace=TRUE)
)
colnames(x) <- c("date", "category", "value")

# creating a cumulative measure making the graphs appear "growing"
library(dplyr)
x <- x %>%
  as_tibble() %>%
  arrange(date) %>%
  mutate(date = as.character(date)) %>%
  group_by(category) %>%
  mutate(cumsum = cumsum(value))

y_max <- max(x$cumsum) + 500

library(doParallel)

all_dates <- unique(x$date)
ncores <- detectCores() - 1
ind_cluster <- sort(rep_len(1:ncores, length(all_dates)))
date_cluster <- split(all_dates, ind_cluster)
registerDoParallel(cl <- makeCluster(ncores))

tmp <- tempfile()

files <- foreach(ic = 1:ncores, .packages = c("tidyverse", "magick")) %dopar% {

  img <- image_graph(1000, 700, res = 96)

  x %>%
    filter(date %in% date_cluster[[ic]]) %>%
    group_by(date) %>%
    do(
      plot = ggplot(.) +
        geom_col(aes(category, cumsum)) +
        scale_y_continuous(expand = c(0, 0), 
                           breaks = seq(0, y_max, 500), 
                           limits = c(0, y_max))
    ) %>%
  pmap(function(date, plot) {
    print(plot + ggtitle(date))
    NULL
  })

  dev.off()

  image_write(image_animate(img, fps = 5), paste0(tmp, ic, ".gif"))
}
stopCluster(cl)

test <- do.call(c, lapply(files, magick::image_read))
test

---

Related Topics