HyperIterator
Caution: this gem monkey patches Ruby's Array
class.
Inspired by Ruby Performance Optimization, HyperIterator is reimplementation of Ruby iterators in Ruby, designed to address performance drawbacks from native implementations, mainly in memory usage.
The main idea is to remove objects from array during iteration. In iteration of large array, this would allow garbage collection to happen before the iteration finishes, therefore reduce memory usage.
Installation
This is in experimental stage. More testing will be conducted. You can see my updates in here
Add this line to your application's Gemfile:
gem 'hyper_iterator'
Or use it with Rails
gem 'hyper_iterator-rails'
And then execute:
$ bundle
Or install it yourself as:
$ gem install hyper_iterator
Available Methods (adding more)
each_slice!
each!
-
HyperIterator.each!
: This is the same aseach!
. But it takes an array as argument. This is added because there are gems that implement their owneach
method (e.g. the pg gem). Instead of monkey patching all the gems out there, this is a compromise for flexibility.
Gotcha
These methods work just as the non bang version, except that, it WILL MUTATE the original array by REMOVING ALL the elements from it.
arr = (1..10).to_a
arr.each! { |el| ... }
# After the iteration, the arr is empty!
# arr => []
Benchmark
For basic benchmarks
rake benchmark
Example report
---------------------------------------------------------
---------------------- each_slice! ----------------------
---------------------------------------------------------
------------------ Garbage Collection -------------------
Array#each_slice
----------------------
100 100 100 100 100 100 100 100 100 100 100 100 100
Array#each_slice!
----------------------
100 92 84 76 68 60 52 44 36 28 20 12 4
---------------------------------------------------------
-------------------- Objects Created --------------------
Array#each_slice
----------------------
# of arrays: 126
# of nodes: 2
Array#each_slice!
----------------------
# of arrays: 125
# of nodes: 0
---------------------------------------------------------
--------------- Execution Time Comparison ---------------
Rehearsal -----------------------------------------------
each_slice 0.040000 0.000000 0.040000 ( 0.049524)
each_slice! 0.040000 0.010000 0.050000 ( 0.047753)
-------------------------------------- total: 0.090000sec
user system total real
each_slice 0.040000 0.010000 0.050000 ( 0.048487)
each_slice! 0.040000 0.010000 0.050000 ( 0.052009)
---------------------------------------------------------
Memory Benchmark
To eliminate the variable of different machines, you will need to install Docker for this.
- Go to Docker to download and install Docker
- Go to the
Dockerfile
, and pick a Ruby version of your choice - Run
bin/setup
in your command line - Run
bin/bm
in your command line
Methodology
The native implementation (non bang version) retains the entire array. So if we keep on adding objects to memory, it will blow up pretty quickly (given the memory is limited).
While using the bang version in this gem, the refernce to the objects store in the array will be removed. So the Garbage Collector (GC) can see them as removable objects and clean them up.
The test code simply creates duplicate object as it iterates. Since the bang version iterator keeps removing the references, the memory gets to recycle itself. In this repot, it will show as more iterations compare to the non bang version.
This is because as the memory gets freed up, we just have more rooms to store new objects, until the GC is not freeing up memory fast enough.
FYI: In this test, the empty array (created from every iteration to store the new objects) is also otaking up more space. If you pop the empty array off (see ./benchmark/memory_bm/each_bang.rb), you can get a lot more iterations!
Example report
- #i is the nth iteration
- The rest is time spent in execution
- The last line is empty because memory blew up
----------------- Array#each ------------------
#i user system total real
0 0.010000 0.000000 0.010000 ( 0.043329)
1 0.000000 0.010000 0.010000 ( 0.010112)
2 0.000000 0.000000 0.000000 ( 0.025059)
3 0.000000 0.000000 0.000000 ( 0.010436)
4 0.000000 0.000000 0.000000 ( 0.008007)
5 0.000000 0.010000 0.010000 ( 0.009692)
6 0.000000 0.010000 0.010000 ( 0.050283)
7
----------------- Array#each! -----------------
#i user system total real
0 0.000000 0.000000 0.000000 ( 0.004570)
1 0.010000 0.000000 0.010000 ( 0.006715)
2 0.000000 0.010000 0.010000 ( 0.040050)
3 0.010000 0.000000 0.010000 ( 0.009338)
4 0.000000 0.000000 0.000000 ( 0.006622)
5 0.000000 0.010000 0.010000 ( 0.019370)
6 0.000000 0.000000 0.000000 ( 0.012097)
7 0.000000 0.010000 0.010000 ( 0.025825)
8 0.000000 0.010000 0.010000 ( 0.014587)
9 0.000000 0.010000 0.010000 ( 0.014893)
10 0.000000 0.000000 0.000000 ( 0.026595)
11 0.000000 0.000000 0.000000 ( 0.012371)
12 0.010000 0.000000 0.010000 ( 0.023964)
13 0.000000 0.000000 0.000000 ( 0.007783)
14 0.000000 0.000000 0.000000 ( 0.004344)
15 0.010000 0.020000 0.030000 ( 0.119632)
16 0.000000 0.010000 0.010000 ( 0.010222)
17 0.000000 0.000000 0.000000 ( 0.009091)
18 %
Development
After checking out the repo, run bin/setup
to install dependencies. Then, run rake test
to run the tests. You can also run bin/console
for an interactive prompt that will allow you to experiment.
Contributing
Bug reports and pull requests are welcome on GitHub at https://github.com/edmundleex/hyper_iterator. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.
TODO
-
each_slice!
should go toEnumerable
module -
each!
forHash
- More comon iterators...
License
The gem is available as open source under the terms of the MIT License.