mirror of
https://github.com/ruby/ruby.git
synced 2026-01-26 20:19:19 +00:00
Since `BUILTIN_TYPE` and `RCLASS_SINGLETON_P` are both stored in `RBasic.flags`, we can combine these two checks in a single bitmask. This rely on `T_ICLASS` and `T_CLASS` not overlapping, and assume `klass` is always either of these types. Just combining the masks brings a small but consistent 1.08x speedup on the simple case benchmark. ``` compare-ruby: ruby 3.5.0dev (2025-08-30T01:45:42Z obj-class 01a57bd6cd) +YJIT +PRISM [arm64-darwin24] built-ruby: ruby 3.5.0dev (2025-08-30T09:56:24Z obj-class 2685f8dbb4) +YJIT +PRISM [arm64-darwin24] | |compare-ruby|built-ruby| |:----------|-----------:|---------:| |obj | 444.410| 478.895| | | -| 1.08x| |extended | 135.139| 140.206| | | -| 1.04x| |singleton | 165.155| 155.832| | | 1.06x| -| |immediate | 380.103| 432.090| | | -| 1.14x| ``` But with the RB_UNLIKELY compiler hint, it's much more significant, however the singleton and enxtended cases are slowed down. However we can assume the simple case is way more common than the other two. ``` compare-ruby: ruby 3.5.0dev (2025-08-30T01:45:42Z obj-class 01a57bd6cd) +YJIT +PRISM [arm64-darwin24] built-ruby: ruby 3.5.0dev (2025-08-30T09:51:01Z obj-class 12d01a1b02) +YJIT +PRISM [arm64-darwin24] | |compare-ruby|built-ruby| |:----------|-----------:|---------:| |obj | 444.951| 556.191| | | -| 1.25x| |extended | 136.836| 113.871| | | 1.20x| -| |singleton | 166.335| 167.747| | | -| 1.01x| |immediate | 379.642| 509.515| | | -| 1.34x| ```
ruby/benchmark
This directory has benchmark definitions to be run with benchmark_driver.gem.
Normal usage
Execute gem install benchmark_driver and run a command like:
# Run a benchmark script with the ruby in the $PATH
benchmark-driver benchmark/app_fib.rb
# Run benchmark scripts with multiple Ruby executables or options
benchmark-driver benchmark/*.rb -e /path/to/ruby -e '/path/to/ruby --jit'
# Or compare Ruby versions managed by rbenv
benchmark-driver benchmark/*.rb --rbenv '2.5.1;2.6.0-preview2 --jit'
# You can collect many metrics in many ways
benchmark-driver benchmark/*.rb --runner memory --output markdown
# Some are defined with YAML for complex setup or accurate measurement
benchmark-driver benchmark/*.yml
See also:
benchmark-driver --help
Usage: benchmark-driver [options] RUBY|YAML...
-r, --runner TYPE Specify runner type: ips, time, memory, once, block (default: ips)
-o, --output TYPE Specify output type: compare, simple, markdown, record, all (default: compare)
-e, --executables EXECS Ruby executables (e1::path1 arg1; e2::path2 arg2;...)
--rbenv VERSIONS Ruby executables in rbenv (x.x.x arg1;y.y.y arg2;...)
--repeat-count NUM Try benchmark NUM times and use the fastest result or the worst memory usage
--repeat-result TYPE Yield "best", "average" or "worst" result with --repeat-count (default: best)
--alternate Alternate executables instead of running the same executable in a row with --repeat-count
--bundler Install and use gems specified in Gemfile
--filter REGEXP Filter out benchmarks with given regexp
--run-duration SECONDS Warmup estimates loop_count to run for this duration (default: 3)
--timeout SECONDS Timeout ruby command execution with timeout(1)
-v, --verbose Verbose mode. Multiple -v options increase visibility (max: 2)
make benchmark
Using make benchmark, make update-benchmark-driver automatically downloads
the supported version of benchmark_driver, and it runs benchmarks with the downloaded
benchmark_driver.
# Run all benchmarks with the ruby in the $PATH and the built ruby
make benchmark
# Or compare with specific ruby binary
make benchmark COMPARE_RUBY="/path/to/ruby --jit"
# Run vm benchmarks
make benchmark ITEM=vm
# Run some limited benchmarks in ITEM-matched files
make benchmark ITEM=vm OPTS=--filter=block
# You can specify the benchmark by an exact filename instead of using the default argument:
# ARGS = $$(find $(srcdir)/benchmark -maxdepth 1 -name '*$(ITEM)*.yml' -o -name '*$(ITEM)*.rb')
make benchmark ARGS=benchmark/erb_render.yml
# You can specify any option via $OPTS
make benchmark OPTS="--help"
# With `make benchmark`, some special runner plugins are available:
# -r peak, -r size, -r total, -r utime, -r stime, -r cutime, -r cstime
make benchmark ITEM=vm_bigarray OPTS="-r peak"