23_Measuring_performance.Rmd

# Measuring performance

**Learning objectives:**

- Understand how to improve your code for making it faster
- Learn what are the tools for improving your code
- Test how to profile your code


## Introduction

> "Before you can make your code faster, you first need to figure out what’s making it slow."


```{r echo=FALSE, fig.align='center',fig.cap="SLOW DOWN TO LEARN HOW TO CODE FASTER | credits: packtpub.com"}
knitr::include_graphics("images/23_code_faster.jpeg")
```


- **profile** your code: measure the run-time of each line of code using realistic inputs
- **experiment** with alternatives to find faster code
- **microbenchmark** to measure the difference in performance.



## Profiling

```{r message=FALSE, warning=FALSE, paged.print=FALSE}
library(profvis)
library(bench)
```


The tool to use is a **profiler**, it allows for **sampling** the code performance through stopping the execution of code every few milliseconds and recording all the steps.

Example:

```{r}
f <- function() {
  pause(0.1)
  g()
  h()
}
g <- function() {
  pause(0.1)
  h()
}
h <- function() {
  pause(0.1)
}
```

Profile the execution of f():

    profvis::pause() is used instead of Sys.sleep()
    profile f(), with utils::Rprof()
    
```{r}
tmp <- tempfile()
Rprof(tmp, interval = 0.1)
f()
Rprof(NULL)
writeLines(readLines(tmp))
```
    
    
**Visualising profiles**

Makes easier to build up a mental model of what you need to change:

    profvis::profvis()
    utils::summaryRprof()

```{r}
source("scripts/profiling-example.R")
profvis(f())
```

**Memory profiling and the garbage collector**

Profiling a loop that modifies an existing variable:
```{r}
profvis::profvis({
  x <- integer()
for (i in 1:1e4) {
  x <- c(x, i)
}
})
```

You can figure out what is the source of the problem by looking at the memory column. In this case, **copy-on-modify** acts in each iteration of the loop creating another copy of x.


**Limitations**

- Profiling does not extend to C code
- Anonymous functions are hard to figure out
- Arguments are evaluated inside another function


### Exercise
```{r eval=FALSE}
profvis::profvis({
  f <- function(n = 1e5) {
  x <- rep(1, n)
  rm(x)
}
},torture = TRUE)
```

    ?rm()
    
[solution](https://advanced-r-solutions.rbind.io/measuring-performance.html)    
    
## Microbenchmarking


*Measurement of the performance of a very small piece of code* is useful for comparing small snippets of code for specific tasks.

```{r echo=FALSE, fig.align='center',fig.cap = "Credits: Google search-engine"}
knitr::include_graphics("images/23_microbenchmarking.jpeg")
```


The {bench} package uses a high precision time.

    bench::mark()
    
    
```{r}
library(bench)
x <- runif(100)
(lb <- bench::mark(
  sqrt(x),
  x ^ 0.5
))
```
- heavily right-skewed distribution


```{r}
require(ggbeeswarm)
plot(lb)
```


## Resources

- [profvis package](https://rstudio.github.io/profvis/)
- [bench package](https://cran.r-project.org/web/packages/bench/bench.pdf)
- [solutions](https://advanced-r-solutions.rbind.io/measuring-performance.html)







## Meeting Videos

### Cohort 1

(no video)

### Cohort 2

`r knitr::include_url("https://www.youtube.com/embed/_zeLDufwTwY")`

### Cohort 3

`r knitr::include_url("https://www.youtube.com/embed/Jdb00nepeWQ")`

### Cohort 4

`r knitr::include_url("https://www.youtube.com/embed/sCso4FAF1DE")`

### Cohort 5

`r knitr::include_url("https://www.youtube.com/embed/pOaiDK7J7EE")`

### Cohort 6

`r knitr::include_url("https://www.youtube.com/embed/d_pzz_AsoRQ")`

### Cohort 7

`r knitr::include_url("https://www.youtube.com/embed/4hngR1c9oP4")`