This post originated from an RSS feed registered with Ruby Buzz
by Alexander Dymo.
Original Post: Ruby Frozen String Literals Do Not Result in Any Performance Improvement
Feed Title: Ruby and Rails Performance Optimization
Feed URL: http://ruby-performance-book.com/blog/feed/
Feed Description: This is the blog about Ruby and Rails performance optimization techniques, tips, and tricks. Learn how the newest Ruby releases perform in the field, see how to reduce memory and CPU usage, discover the world of profiling, measuring, and performance testing.
Frozen string literals is the experimental feature that appeared in Ruby 2.3 and will become the default behavior in Ruby 3.0. There's been some talk about how that improves performance. A month ago I ported one of my applications and saw no improvement. Today took another look, and my conclusion still stands.
If you haven't heard about frozen string literals yet, please read my previous blog post to learn more about how they work.
When I last experimented with this feature, my conclusion was that it only adds to the developer's frustration. It's just more work for us to port our applications.
I finally found the time to port Acunote, the application that has decent performance benchmarks. Let's look at some of my before/after measurements:
Benchmark
Default Behavior
Frozen String Literals
Benchmark Burndown 120
0.91 ± 0.02
0.91 ± 0.02
Benchmark Sprint 20 x (1+5) (C)
0.25 ± 0.00
0.25 ± 0.01
Benchmark Analyze 20 x (1+5)
1.96 ± 0.02
1.95 ± 0.04
Benchmark Task Details Page
0.40 ± 0.01
0.41 ± 0.01
Benchmark Signup
0.30 ± 0.00
0.31 ± 0.01
Benchmark Copy 120
1.47 ± 0.01
1.51 ± 0.02
Benchmark Import 71
2.02 ± 0.02
2.10 ± 0.01
Benchmark Set Field (EP)
0.20 ± 0.00
0.21 ± 0.00
This table is incomplete. The whole performance test suite has more than 60 benchmarks. But not one of them indicated any statistically significant speedup. Worse, the performance actually declined in two of the benchmarks that I highlighted in the table.
I haven't investigated the causes of the slowdown yet, but the conclusion is clear. Frozen string literals do not improve performance of my Rails applications.
The word my is important of course. There might well be a valid case when frozen string literals make your code faster. I just haven't seen it yet. But I have hard times imagining what it might be.
Did you like this post? Follow me on
Twitter
or
Google+
to stay updated on Ruby performance optimization news.
Previous Topic
