profiler.pdf

(111 KB) Pobierz

Profiling R Code

Roger D. Peng, Associate Professor of Biostatistics

Johns Hopkins Bloomberg School of Public Health

Why is My Code So Slow?

Proﬁling is a systematic way to examine how much time is spend in different parts of a program

Useful when trying to optimize your code

Often code runs ﬁne once, but what if you have to put it in a loop for 1,000 iterations? Is it still fast

enough?

Proﬁling is better than guessing

2/17

On Optimizing Your Code

Getting biggest impact on speeding up code depends on knowing where the code spends most

of its time

This cannot be done without performance analysis or proﬁling

We should forget about small efficiencies, say about 97% of the time: premature

optimization is the root of all evil

--Donald Knuth

3/17

General Principles of Optimization

Design ﬁrst, then optimize

Remember: Premature optimization is the root of all evil

Measure (collect data), don’t guess.

If you're going to be scientist, you need to apply the same principles here!

4/17

Using

system.time()

Takes an arbitrary R expression as input (can be wrapped in curly braces) and returns the

amount of time taken to evaluate the expression

Computes the time (in seconds) needed to execute an expression

If there’s an error, gives time until the error occurred

Returns an object of class

proc_time

- user time:

time charged to the CPU(s) for this expression

- elapsed time:

"wall clock" time

5/17