Add parse benchmark #104
Add parse benchmark #104
Conversation
naitoh
force-pushed
the
add_parse_benchmark
branch
from
1ba69ad
to
31334b5
Compare
naitoh
force-pushed
the
add_parse_benchmark
branch
from
31334b5
to
55a145b
Compare
Rakefile
Outdated
| benchmark_tasks << "benchmark:#{name}" | ||
|
|
||
| case name | ||
| when /\Aparse/, "shift" |
There was a problem hiding this comment.
| when /\Aparse/, "shift" | |
| when /\Aparse/ |
benchmark/parse.yaml
Outdated
| 'dom' : REXML::Document.new(xml).elements.each("root/child") {|_|} | ||
| 'sax' : REXML::Parsers::SAX2Parser.new(xml).parse | ||
| 'pull' : | | ||
| p = REXML::Parsers::PullParser.new(xml) |
There was a problem hiding this comment.
How about using parser for variable name because we have p method?
| p = REXML::Parsers::PullParser.new(xml) | |
| parser = REXML::Parsers::PullParser.new(xml) |
benchmark/parse.yaml
Outdated
| 'pull' : | | ||
| p = REXML::Parsers::PullParser.new(xml) | ||
| while p.has_next? | ||
| res = p.pull |
There was a problem hiding this comment.
It seems that we don't need to assign the return value:
| res = p.pull | |
| p.pull |
benchmark/parse.yaml
Outdated
| n_elements = Integer(ENV.fetch("N_ELEMENTS", "10000"), 10) | ||
| n_attributes = Integer(ENV.fetch("N_ATTRIBUTES", "10"), 10) |
There was a problem hiding this comment.
With these configurations, we take 10+ minutes for one rake benchmark: https://github.com/ruby/rexml/actions/runs/7431570516/job/20222471649?pr=104
Can we reduce benchmark time? Or do we need these numbers for meaningful benchmark?
There was a problem hiding this comment.
I set this value because I need to execute XML with a large number of elements, but the execution time is too long, so I will reduce the number of elements.
There was a problem hiding this comment.
https://github.com/ruby/rexml/actions/runs/7434389228/job/20228488163?pr=104
Reduced to 1/4. (2 minutes 38 seconds)
naitoh
force-pushed
the
add_parse_benchmark
branch
from
55a145b
to
e4af9f3
Compare
|
Thanks! |
I want to improve the parsing process and would like to add a parsing benchmark.
The benchmark process just parses the XML from beginning to end.
Since performance differs depending on whether YJIT is ON or OFF, both are measured.