Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/tbrowder/raku-read-write-tests
https://github.com/tbrowder/raku-read-write-tests
Last synced: 23 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/tbrowder/raku-read-write-tests
- Owner: tbrowder
- License: artistic-2.0
- Created: 2016-03-31T15:59:48.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-10-06T21:29:03.000Z (over 7 years ago)
- Last Synced: 2024-12-09T19:50:19.248Z (about 1 month ago)
- Language: Perl 6
- Size: 146 KB
- Stars: 4
- Watchers: 3
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE
Awesome Lists containing this project
README
# Perl 6 Read/Write Tests
[](https://travis-ci.org/tbrowder/perl6-read-write-tests)
Perl 6 is currently slow compared to Perl 5 reading ASCII text
files line by line (although the time differences are not so large
when UTF-8 files are read). Such line processing is a staple of data
processing in use cases such as analyzing output of many kinds of
programs. An example is the category of programs known as raytracing,
one of which, used heavily by analysts and scientists, is
[BRL-CAD](http://brlcad.org).
This test suite was developed to monitor the progress of Perl 6 versus
Perl 5 in closing the gap of processing an ASCII or UTF-8 file of many lines.
The test files are of varying numbers of lines, each line consisting
of 100 bytes, including the ending newline.
The tests currently compare Perl 6 versus Perl 5 under the following conditions:
+ reading UTF-8 files (Perl 6 native read, Perl 5 using the '<:encoding(UTF-8)' form of open)
+ reading ASCII files (native read with both Perl versions)
+ reading ASCII files (Perl 6 using the ':enc' form of open)
+ reading ASCII files (Perl 6 using the ':enc' form of open)
## Running the tests
1. Edit file `run-rw-tests.p6` to set the value of array `@S` to the
number and size of tests to run. Note that a bare number will be
interpreted as Megabytes and a number with an 'M' or 'G' appended
will be interpreted accordingly. The settings that generate most
test logs shown on the github site are:
@S = <1m 1g 5g 10g>; # 10K, 10M, 50M, 100M lines, respectively
$ntrials = 3;
2. Edit file `run-rw-tests.p6` to set the variable `$ntrials` for
the number of trials desired for each size file
3. Exexute file `run-rw-tests.p6`. Two log files will be generated and
each placed in the appropriate `logs` subdirectories (which will be
created if they don't exist).
## Results of recent file read tests
### Before latest IO improvements
+ Date: 2017-05-17
+ File type: ASCII
+ Rakudo version: 2017.04.3-275-g84502dc
+ Perl 5 version: 5.20.2
| Size | Number lines | Trials | Perl 5 T | Perl 6 T | P6T / P5T |
| ---: | ---: | :---: | ---: | ---: | :---: |
| 1 Mb | 10,000 | 3 | 0.03 s | 0.86 s | 28.7 |
| 1 Gb | 10,000,000 | 3 | 2.73 s | 51.50 s | 18.9 |
| 5 Gb | 50,000,000 | 3 | 27.41 s | 410.00 s | 15.0 |
| 10 Gb | 100,000,000 | 3 | 66.39 s | 860.56 s | 13.0 |
### After latest IO improvements
+ Date: 2017-06-29
+ File type: ASCII
+ Rakudo version: 2017.06-45-g86e7b2b
+ Perl 5 version: 5.20.2
| Size | Number lines) | Trials | Perl 5 T | Perl 6 T | P6T / P5T | Improvement
| ---: | ---: | :---: | ---: | ---: | :---: | :--: |
| 1 Mb | 10,000 | 3 | 0.03 s | 0.60 s | 20.0 | 30.3% |
| 1 Gb | 10,000,000 | 3 | 2.67 s | 32.27 s | 12.1 | 36.0% |
| 5 Gb | 50,000,000 | 3 | 24.37 s | 276.41 s | 11.3 | 24.7% |
| 10 Gb | 100,000,000 | 3 | 55.40 s | 582.37 s | 10.5 | 19.2%|
Notes:
1. See the complete results in log files in the **logs** subdirectories.
2. **T** - Time: user time from the GNU `time` program.
3. **Improvement** - (prev P6T/P5T less curr P6T/P5T)/(prev P6T/P5T)
4. When the number of trials is > 1, the user time data are averages over that number.
5. This suite is developed on a Debian system with no consideration to
it running successfully on any other OS. Pull requests are
welcome.