* Rewrite Kernel#tap with Ruby
This was good for VM too, but of course my intention is to unblock JIT's inlining of a block over yield
(inlining invokeyield has not been committed though).
* Fix test_settracefunc
About the :tap deletions, the :tap events are actually traced (we already have a TracePoint test for builtin methods),
but it's filtered out by tp.path == "xyzzy" (it became "<internal:kernel>"). We could trace tp.path == "<internal:kernel>"
cases too, but the lineno is impacted by kernel.rb changes and I didn't want to make it fragile for kernel.rb lineno changes.
for opt_* insns.
opt_eq handles rb_obj_equal inside opt_eq, and all other cfunc is
handled by opt_send_without_block. Therefore we can't decide which insn
should be generated by checking whether it's cfunc cc or not.
```
$ benchmark-driver -v --rbenv 'before --jit;after --jit' benchmark/mjit_opt_cc_insns.yml --repeat-count=4
before --jit: ruby 2.8.0dev (2020-06-26T05:21:43Z master 9dbc2294a6) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-06-26T06:30:18Z master 75cece1b0b) +JIT [x86_64-linux]
last_commit=Decide JIT-ed insn based on cached cfunc
Calculating -------------------------------------
before --jit after --jit
mjit_nil?(1) 73.878M 74.021M i/s - 40.000M times in 0.541432s 0.540391s
mjit_not(1) 72.635M 74.601M i/s - 40.000M times in 0.550702s 0.536187s
mjit_eq(1, nil) 7.331M 7.445M i/s - 8.000M times in 1.091211s 1.074596s
mjit_eq(nil, 1) 49.450M 64.711M i/s - 8.000M times in 0.161781s 0.123627s
Comparison:
mjit_nil?(1)
after --jit: 74020528.4 i/s
before --jit: 73878185.9 i/s - 1.00x slower
mjit_not(1)
after --jit: 74600882.0 i/s
before --jit: 72634507.6 i/s - 1.03x slower
mjit_eq(1, nil)
after --jit: 7444657.4 i/s
before --jit: 7331304.3 i/s - 1.02x slower
mjit_eq(nil, 1)
after --jit: 64710790.6 i/s
before --jit: 49449507.4 i/s - 1.31x slower
```
because opt_nil/opt_not/opt_eq populates cc even when it doesn't
fallback to opt_send_without_block because of vm_method_cfunc_is.
```
$ benchmark-driver -v --rbenv 'before --jit;after --jit' benchmark/mjit_opt_cc_insns.yml --repeat-count=4
before --jit: ruby 2.8.0dev (2020-06-22T08:11:24Z master d231b8f95b) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-06-22T08:53:27Z master e1125879ed) +JIT [x86_64-linux]
last_commit=Compile opt_send for opt_* only when cc has ISeq
Calculating -------------------------------------
before --jit after --jit
mjit_nil?(1) 54.106M 73.693M i/s - 40.000M times in 0.739288s 0.542795s
mjit_not(1) 53.398M 74.477M i/s - 40.000M times in 0.749090s 0.537075s
mjit_eq(1, nil) 7.427M 6.497M i/s - 8.000M times in 1.077136s 1.231326s
Comparison:
mjit_nil?(1)
after --jit: 73692594.3 i/s
before --jit: 54106108.4 i/s - 1.36x slower
mjit_not(1)
after --jit: 74477487.9 i/s
before --jit: 53398125.0 i/s - 1.39x slower
mjit_eq(1, nil)
before --jit: 7427105.9 i/s
after --jit: 6497063.0 i/s - 1.14x slower
```
Actually opt_eq becomes slower by this. Maybe it's indeed using
opt_send_without_block, but I'll approach that one in another commit.
These days I don't use `make benchmark`. The YAML files should be
executable with bare `benchmark-driver` CLI without passing
`RUBYOPT=-Ibenchmark/lib`.
A prerequisite to fix https://bugs.ruby-lang.org/issues/15589 with JIT.
This commit alone doesn't make a significant difference yet, but I thought
this commit should be committed independently.
This method override was discussed in [Misc #16961].
The vm1_ prefix and vm2_ had had special meaning until
820ad9cb1d and
12068aa4e9. AFAIK there's no special
meaning in vm3_ prefix.
As they have confused people (like "In `benchmark` what is difference
between `vm1_`, `vm2_` and `vm3_`"), I'd like to remove the obsoleted
prefix as we obviated that two years ago.
for VM_METHOD_TYPE_CFUNC.
This has been known to decrease optcarrot fps:
```
$ benchmark-driver -v --rbenv 'before --jit;after --jit' benchmark.yml --repeat-count=24 --output=all
before --jit: ruby 2.8.0dev (2020-04-13T16:25:13Z master fb40495cd9) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-04-13T23:23:11Z mjit-inline-c bdcd06d159) +JIT [x86_64-linux]
Calculating -------------------------------------
before --jit after --jit
Optcarrot Lan_Master.nes 66.38132676191719 67.41369177299630 fps
69.42728743772243 68.90327567263054
72.16028300263211 69.62605130880686
72.46631319102777 70.48818243767207
73.37078877002490 70.79522887347566
73.69422431217367 70.99021920193194
74.01471487018695 74.69931965402584
75.48685183295630 74.86714575949016
75.54445264507932 75.97864419721677
77.28089738169756 76.48908637569581
78.04183397891302 76.54320932488021
78.36807984096562 76.59407262898067
78.92898762543574 77.31316743361343
78.93576483233765 77.97153484180480
79.13754917503078 77.98478782102325
79.62648945850653 78.02263322726446
79.86334213878064 78.26333724045934
80.05100635898518 78.60056756355614
80.26186843769584 78.91082645644468
80.34205717020330 79.01226659142263
80.62286066044338 79.32733939423721
80.95883033058557 79.63793060542024
80.97376819251613 79.73108936622778
81.23050939202896 80.18280109433088
```
and I deleted this capability in an early stage of YARV-MJIT development:
0ab130feee
I suspect either of the following things could be the cause:
* Directly calling vm_call_cfunc requires more optimization effort in GCC,
resulting in 30ms-ish compilation time increase for such methods and
decreasing the number of methods compiled in a benchmarked period.
* Code size increase => icache miss hit
These hypotheses could be verified by some methodologies. However, I'd
like to introduce this regardless of the result because this blocks
inlining C method's definition.
I may revert this commit when I give up to implement inlining C method
definition, which requires this change.
Microbenchmark-wise, this gives slight performance improvement:
```
$ benchmark-driver -v --rbenv 'before --jit;after --jit' benchmark/mjit_send_cfunc.yml --repeat-count=4
before --jit: ruby 2.8.0dev (2020-04-13T16:25:13Z master fb40495cd9) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-04-13T23:23:11Z mjit-inline-c bdcd06d159) +JIT [x86_64-linux]
Calculating -------------------------------------
before --jit after --jit
mjit_send_cfunc 41.961M 56.489M i/s - 100.000M times in 2.383143s 1.770244s
Comparison:
mjit_send_cfunc
after --jit: 56489372.5 i/s
before --jit: 41961388.1 i/s - 1.35x slower
```
Previously, passing a keyword splat to a method always allocated
a hash on the caller side, and accepting arbitrary keywords in
a method allocated a separate hash on the callee side. Passing
explicit keywords to a method that accepted a keyword splat
did not allocate a hash on the caller side, but resulted in two
hashes allocated on the callee side.
This commit makes passing a single keyword splat to a method not
allocate a hash on the caller side. Passing multiple keyword
splats or a mix of explicit keywords and a keyword splat still
generates a hash on the caller side. On the callee side,
if arbitrary keywords are not accepted, it does not allocate a
hash. If arbitrary keywords are accepted, it will allocate a
hash, but this commit uses a callinfo flag to indicate whether
the caller already allocated a hash, and if so, the callee can
use the passed hash without duplicating it. So this commit
should make it so that a maximum of a single hash is allocated
during method calls.
To set the callinfo flag appropriately, method call argument
compilation checks if only a single keyword splat is given.
If only one keyword splat is given, the VM_CALL_KW_SPLAT_MUT
callinfo flag is not set, since in that case the keyword
splat is passed directly and not mutable. If more than one
splat is used, a new hash needs to be generated on the caller
side, and in that case the callinfo flag is set, indicating
the keyword splat is mutable by the callee.
In compile_hash, used for both hash and keyword argument
compilation, if compiling keyword arguments and only a
single keyword splat is used, pass the argument directly.
On the caller side, in vm_args.c, the callinfo flag needs to
be recognized and handled. Because the keyword splat
argument may not be a hash, it needs to be converted to a
hash first if not. Then, unless the callinfo flag is set,
the hash needs to be duplicated. The temporary copy of the
callinfo flag, kw_flag, is updated if a hash was duplicated,
to prevent the need to duplicate it again. If we are
converting to a hash or duplicating a hash, we need to update
the argument array, which can including duplicating the
positional splat array if one was passed. CALLER_SETUP_ARG
and a couple other places needs to be modified to handle
similar issues for other types of calls.
This includes fairly comprehensive tests for different ways
keywords are handled internally, checking that you get equal
results but that keyword splats on the caller side result in
distinct objects for keyword rest parameters.
Included are benchmarks for keyword argument calls.
Brief results when compiled without optimization:
def kw(a: 1) a end
def kws(**kw) kw end
h = {a: 1}
kw(a: 1) # about same
kw(**h) # 2.37x faster
kws(a: 1) # 1.30x faster
kws(**h) # 2.19x faster
kw(a: 1, **h) # 1.03x slower
kw(**h, **h) # about same
kws(a: 1, **h) # 1.16x faster
kws(**h, **h) # 1.14x faster
Instead of searching twice to extract and to delete, extract and
delete the found position at the first search.
This makes faster nearly twice, for regexps and strings.
| |compare-ruby|built-ruby|
|:-------------|-----------:|---------:|
|regexp-short | 2.143M| 3.918M|
|regexp-long | 105.162k| 205.410k|
|string-short | 3.789M| 7.964M|
|string-long | 1.301M| 2.457M|
I noticed that some files in rubygems were executable, and I could think
of no reason why they should be.
In general, I think ruby files should never have the executable bit set
unless they include a shebang, so I run the following command over the
whole repo:
```bash
find . -name '*.rb' -type f -executable -exec bash -c 'grep -L "^#!" $1 || chmod -x $1' _ {} \;
```
* Stop making a redundant hash copy in Hash#dup
It was making a copy of the hash without rehashing, then created an
extra copy of the hash to do the rehashing. Since rehashing creates
a new copy already, this change just uses that rehashing to make
the copy.
[Bug #16121]
* Remove redundant Check_Type after to_hash
* Fix freeing and clearing destination hash in Hash#initialize_copy
The code was assuming the state of the destination hash based on the
source hash for clearing any existing table on it. If these don't match,
then that can cause the old table to be leaked. This can be seen by
compiling hash.c with `#define HASH_DEBUG 1` and running the following
script, which will crash from a debug assertion.
```ruby
h = 9.times.map { |i| [i, i] }.to_h
h.send(:initialize_copy, {})
```
* Remove dead code paths in rb_hash_initialize_copy
Given that `RHASH_ST_TABLE_P(h)` is defined as `(!RHASH_AR_TABLE_P(h))`
it shouldn't be possible for a hash to be neither of these, so there
is no need for the removed `else if` blocks.
* Share implementation between Hash#replace and Hash#initialize_copy
This also fixes key rehashing for small hashes backed by an array
table for Hash#replace. This used to be done consistently in ruby
2.5.x, but stopped being done for small arrays in ruby 2.6.x.
This also bring optimization improvements that were done for
Hash#initialize_copy to Hash#replace.
* Add the Hash#dup benchmark
I noticed that in case of cache misshit, re-calculated cc->me can
be the same method entry than the pevious one. That is an okay
situation but can't we partially reuse the cache, because cc->call
should still be valid then?
One thing that has to be special-cased is when the method entry
gets amended by some refinements. That happens behind-the-scene
of call cache mechanism. We have to check if cc->me->def points to
the previously saved one.
Calculating -------------------------------------
trunk ours
vm2_poly_same_method 1.534M 2.025M i/s - 6.000M times in 3.910203s 2.962752s
Comparison:
vm2_poly_same_method
ours: 2025143.9 i/s
trunk: 1534447.2 i/s - 1.32x slower
This approach is simpler than the previous approach which tries to
emulate realpath(3). It also performs much better on both Linux and
OpenBSD on the included benchmarks.
By using realpath(3), we can better integrate with system security
features such as OpenBSD's unveil(2) system call.
This does not use realpath(3) on Windows even if it exists, as the
approach for checking for absolute paths does not work for drive
letters. This can be fixed without too much difficultly, though until
Windows defines realpath(3), there is no need to do so.
For File.realdirpath, where the last element of the path is not
required to exist, fallback to the previous approach, as realpath(3)
on most operating systems requires the whole path be valid (per POSIX),
and the operating systems where this isn't true either plan to conform
to POSIX or may change to conform to POSIX in the future.
glibc realpath(3) does not handle /path/to/file.rb/../other_file.rb
paths, returning ENOTDIR in that case. Fallback to the previous code
if realpath(3) returns ENOTDIR.
glibc doesn't like realpath(3) usage for paths like /dev/fd/5,
returning ENOENT even though the path may appear to exist in the
filesystem. If ENOENT is returned and the path exists, then fall
back to the default approach.
and switch-case branches.
Buffer allocation optimization using `ALLOCA_N` would be the main
benefit of patch. It eliminates the O(N) buffer extensions.
It also reduces the number of branches using escape table like
https://mattn.kaoriya.net/software/lang/c/20160817011915.htm.
Closes: https://github.com/ruby/ruby/pull/2226
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
Co-authored-by: Yasuhiro MATSUMOTO <mattn.jp@gmail.com>
and switch-case branches.
Buffer allocation optimization using `ALLOCA_N` would be the main
benefit of patch. It eliminates the O(N) buffer extensions.
It also reduces the number of branches using escape table like
https://mattn.kaoriya.net/software/lang/c/20160817011915.htm.
Closes: https://github.com/ruby/ruby/pull/2226
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
Co-authored-by: Yasuhiro MATSUMOTO <mattn.jp@gmail.com>
I heard actually this part would not be a bottleneck for rendering
because writing anything to terminal takes way longer time anyway, but I
thought this benchmark script might be useful for benchmarking Ruby
itself.
To prevent noise for benchmark result. Just for the case.
[Bug #15552]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66893 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
To support the change of default encoding.
It had not worked correctly since 2.0 :-)
[Bug #15552]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66892 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Add new benchmark scripts for binary operations of Complex with float
components.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66680 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* benchmark/lib/load.rb: add small utility which requires
benchmark-driver.rb. You can load this file and can
use benchmark-driver.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64870 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Simply use DISPATCH_ORIGINAL_INSN instead of rb_funcall. This is,
when possible, overall performant because method dispatch results are
cached inside of CALL_CACHE. Should also be good for JIT.
----
trunk: ruby 2.6.0dev (2018-09-12 trunk 64689) [x86_64-darwin15]
ours: ruby 2.6.0dev (2018-09-12 leaf-insn 64688) [x86_64-darwin15]
last_commit=make opt_str_freeze leaf
Calculating -------------------------------------
trunk ours
vm2_freezestring 5.440M 31.411M i/s - 6.000M times in 1.102968s 0.191017s
Comparison:
vm2_freezestring
ours: 31410864.5 i/s
trunk: 5439865.4 i/s - 5.77x slower
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64690 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
because it's only available for limited platforms for now.
I'll make it portable and show it later.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63954 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark/README.md: fix help output, which is changed on v0.14.6.
Especially `e1::path1,arg1,...; e2::path2,arg2` part was wrong since `,`
can't be used to split arguments anymore.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63945 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
to improve the accuracy of measurement by stop using block.
benchmark/app_erb.rb -> benchmark/app_erb.yml: renamed and revised
benchmark/erb_render.rb -> benchmark/erb_render.yml: ditto
benchmark/README.md: follow renames
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63941 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
the original behavior of benchmark/driver.rb.
Probably I won't use this but this is requested by ko1.
Use this with:
make benchmark OPTS="-o driver"
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63937 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
supported by legacy benchmark/driver.rb.
benchmark/README.md: document them
common.mk: update benchmark_driver to correct 0.0 output and to fix
spacing format of `-o simple` and `-o markdown`.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63933 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
runner/peak.rb: ditto
This is needed to make commands like `make -C .ruby-svn benchmark
ITEM=erb OPTS="-r size -o simple"` succeed.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63932 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark_driver is the official way to install benchmark_driver.
benchmark-driver is just an alias for it.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63930 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark/*.rb is only benchmarks now. We don't need prefixes.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63928 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This reverts r63900.
Having single-execution benchmark as a normal Ruby script is preferred
by ko1. I'm not a big fan of having inconsistent benchmark formats, but
I can understand some benefits of it.
common.mk: remove obsolsted benchmark-each PHONY declaration, support
running Ruby scripts added by this commit.
README.md: follow ARGS change
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63926 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
by adding runner plugins for them.
benchmark/lib/benchmark_driver/runner/peak.rb: added peak runner plugin
benchmark/lib/benchmark_driver/runner/size.rb: added size runner plugin
common.mk: allow using them
benchmark/memory_wrapper.rb: deleted in favor of those runner plugins
benchmark/README.md: document them
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63924 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
in favor of just using benchmark_driver.gem.
common.mk: The new `make benchmark` covers the both usages for old `make
benchmark` and old `make benchmark-each`. So `make benchmark-each` is
dropped now.
benchmark/README.md: Explain its details
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63918 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark/driver.rb: deal with breaking changes which are actually
introduced for this driver.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63916 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
to original benchmark-driver command.
I'm going to add some runner plugins to resurrect metrics which were
originally supported by benchmark/driver.rb...
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63906 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Now all benchmarks are converted to YAMLs.
common.mk: Drop obsoleted bm_* pattern
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63902 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
scripts.
This is needed to finish converting Ruby scripts to YAMLs.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63899 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
for now.
When measured script is really too fast, while loop substituion may
return a negative benchmark result.
Probably benchmark_driver.gem has rooms to be improved about this.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63895 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark/driver.rb had removed the cost for while loop in benchmark/bm_vm1_*.rb,
and benchmark_driver.gem can achieve the same thing with `loop_count`.
But unfortunately current benchmark_driver.gem can't solve it only for vm1_yield.yml...
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63893 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This YAML transformation is needed to support whileloop2 time substituion
by benchmark_driver.gem later.
This commmit changes no benchmark behavior.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63892 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This YAML transformation is needed to support whileloop time substituion
by benchmark_driver.gem later.
This commmit changes no benchmark behavior.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63891 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Makefile.in: Clone benchmark-driver repository in benchmark/benchmark-driver
`make update-benchmark-driver`, like simplecov.
win32/Makefile.sub: Roughly do the same thing.
.gitignore: Ignore the cloned repository.
common.mk: Trigger `make update-benchmark-driver` to run `make benchmark`
and adjust arguments for benchmark_driver.gem.
benchmark/require.yml: renamed from benchmark/bm_require.rb, benchmark/prepare_require.rb
benchmark/require_thread.yml: renamed from benchmark/bm_require_thread.rb, benchmark/prepare_require_thread.rb
benchmark/so_count_words.yml: renamed from benchmark/bm_so_count_words.rb, benchmark/prepare_so_count_words.rb,
benchmark/wc.input.base
benchmark/so_k_nucleotide.yml: renamed from benchmark/bm_so_k_nucleotide.rb, benchmark/prepare_so_k_nucleotide.rb,
benchmark/make_fasta_output.rb
benchmark/so_reverse_complement.yml: renamed from benchmark/bm_so_reverse_complement.rb, benchmark/prepare_so_reverse_complement.rb,
benchmark/make_fasta_output.rb
I'm sorry but I made some duplications between benchmark/require.yml and benchmark/require_thread.yml,
and between benchmark/so_k_nucleotide.yml and benchmark/so_reverse_complement.yml.
If you're not comfortable with it, please combine these YAMLs to share
the same prelude. One YAML file can have multiple benchmark definitions
sharing prelude.
benchmark/driver.rb: Replace its core feature with benchmark_driver.gem.
Some old features are gone for now, but I'll add them again later.
[Misc #14902]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63888 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
It seems like they are all benchmark drivers but "benchmark/driver.rb"
is the latest and others are no longer used.
It's confusing to have multiple drivers (and actually I used
benchmark/run.rb since I didn't know I should use benchmark/driver.rb).
As I'm going to support only benchmark/driver.rb features in Misc#14902,
let me delete them.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63886 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
"Real" time is too unstable on my systems, hopefully counting
only CPU time can gain more reliable benchmark results.
[ruby-core:87362] [Feature #14815]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63564 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
benchmark/memory_wrapper.rb will Kernel#load these
scripts, preventing DATA from being initialized, so
use heredoc instead.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@63497 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* benchmark/driver.rb: use `--version` instead of `-v` to get version
information.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62515 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
[Feature #14045]
* insns.def (getblockparam, setblockparam): add special access
instructions for block parameters.
getblockparam checks VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM and
if it is not set this instruction creates a Proc object from
a given blcok and set VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM.
setblockparam is similar to setlocal, but set
VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM.
* compile.c: use get/setblockparm instead get/setlocal instructions.
Note that they are used for method local block parameters (def m(&b)),
not for block local method parameters (iter{|&b|).
* proc.c (get_local_variable_ptr): creates Proc object for
Binding#local_variable_get/set.
* safe.c (safe_setter): we need to create Proc objects for postponed
block parameters when $SAFE is changed.
* vm_args.c (args_setup_block_parameter): used only for block local blcok
parameters.
* vm_args.c (vm_caller_setup_arg_block): if called with
VM_CALL_ARGS_BLOCKARG_BLOCKPARAM flag then passed block values should be
a block handler.
* test/ruby/test_optimization.rb: add tests.
* benchmark/bm_vm1_blockparam*: added.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60397 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This means File.chmod, File.lchmod, File.chown, File.lchown,
File.unlink, and File.utime operations on slow filesystems
no longer hold up other threads.
The platform-specific utime_failed changes is compile-tested
using a new UTIME_EINVAL macro
This hurts performance on fast filesystem, but these methods
are unlikely to be performance bottlenecks and (IMHO) avoiding
pathological slowdowns and stalls are more important.
benchmark results:
minimum results in each 3 measurements.
Execution time (sec)
name trunk built
file_chmod 0.591 0.801
Speedup ratio: compare with the result of `trunk' (greater is better)
name built
file_chmod 0.737
* file.c (UTIME_EINVAL): new macro to ease compile-testing
* file.c (struct apply_arg): new struct
* file.c (no_gvl_apply2files): new function
* file.c (apply2files): release GVL
* file.c (chmod_internal): adjust for apply2files changes
* file.c (lchmod_internal): ditto
* file.c (chown_internal): ditto
* file.c (lchown_internal): ditto
* file.c (utime_failed): ditto
* file.c (utime_internal): ditto
* file.c (unlink_internal): ditto
[ruby-core:83200] [Feature #13996]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60386 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This converts all slow syscalls in the Dir.empty? implementation
to release GVL. We avoid unnecessarily GVL release and
reacquire for each slow call (opendir, readdir, closedir) and
instead only release and acquire the GVL once in the common
case.
Benchmark results show a small degradation in single-threaded
performance:
Execution time (sec)
name trunk built
dir_empty_p 0.689 0.758
Speedup ratio: compare with the result of `trunk' (greater is better)
name built
dir_empty_p 0.909
* dir.c (rb_gc_for_fd_with_gvl): new function
(nogvl_dir_empty_p): ditto
(dir_s_empty_p): use new functions to release GVL
* benchmark/bm_dir_empty_p.rb: new benchmark
[ruby-core:83071] [Feature #13958]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60111 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
rename(2) requires two pathname resolution operations which can
take considerable time on slow filesystems, release the GVL so
operations on other threads may proceed.
On fast, local filesystems, this change results in some slowdown
as shown by the new benchmark. I consider the performance trade
off acceptable as cases are avoided.
benchmark results:
minimum results in each 3 measurements.
Execution time (sec)
name trunk built
file_rename 2.648 2.804
Speedup ratio: compare with the result of `trunk' (greater is better)
name built
file_rename 0.944
* file.c (no_gvl_rename): new function
(rb_file_s_rename): release GVL for renames
* benchmark/bm_file_rename.rb: new benchmark
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60088 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* string.c (rb_strseq_index): refactor and avoid
call of str_strlen() when offset == 0.
it will improve performance of String#index and #include?
* benchmark/bm_string_index.rb: benchmark for this change
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60086 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
[Feature #13884]
Reduce number of memory allocations for "and", "or" and "diff"
operations on small arrays
Very often, arrays are used to filter parameters and to select
interesting items from 2 collections and very often these
collections are small enough, for example:
```ruby
SAFE_COLUMNS = [:id, :title, :created_at]
def columns
@all_columns & SAFE_COLUMNS
end
```
In this patch, I got rid of unnecessary memory allocations for
small arrays when "and", "or" and "diff" operations are performed.
name | HEAD | PATCH
-----------------+------:+------:
array_small_and | 0.615 | 0.263
array_small_diff | 0.676 | 0.282
array_small_or | 0.953 | 0.463
name | PATCH
-----------------+------:
array_small_and | 2.343
array_small_diff | 2.392
array_small_or | 2.056
name | HEAD | PATCH
-----------------+------:+------:
array_small_and | 1.429 | 1.005
array_small_diff | 1.493 | 0.878
array_small_or | 1.672 | 1.152
name | PATCH
-----------------+------:
array_small_and | 1.422
array_small_diff | 1.700
array_small_or | 1.452
Author: Dmitry Bochkarev <dimabochkarev@gmail.com>
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60057 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
from bm_app_erb_render.rb.
I'm told from ko1 that bm_app_* is namespace for Ruby applications,
not for ERB and we should use bm_erb_* for ERB benchmark instead.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58915 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
to skip object allocation for static string.
We can't always enable frozen_string_literal pragma because we can't
freeze string literals embedded by user for backward compatibility.
So we need to use fstring for each static string.
Since adding ".freeze" to string literals in #content_dump is slow
on compiling, I used unary "-" operator instead.
benchmark/bm_app_erb_render.rb: Added rendering-only benchmark to
test rendering performance on production environment.
This benchmark is created to reproduce the behavior on Sinatra (Tilt).
Thus it doesn't use ERB#result to skip parsing compiled code.
It doesn't use ERB#def_method too to regard `title` and `content` as
local variables. If we use #def_method, `title` and `content` needs
to be method call. I wanted to avoid it.
This patch's benchmark results is:
* Before
app_erb_render 1.250
app_erb 0.704
* After
app_erb_render 1.066
app_erb 0.686
This patch optimizes rendering performance (app_erb_render) without
spoiling (total of rendering +) compiling performance (app_erb).
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58905 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
There are currently no benchmarks for Fiber performance, I
should've committed this years ago when [Feature #10341] was
implemented.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58606 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
I was about to write off this benchmark while working on GVL
improvements on multi-core systems.
However I noticed it exposes a weakness in my work-in-progress
code when I tested on an old single CPU system. Further testing
reveals setting CPU affinity ("schedtool -a 0x1" on Linux) on a
modern multi-core system is enough to reproduce the problem
exposed by this benchmark.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58571 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
"lazy_sweep" does not appear to have ever been a valid kwarg
for GC.start, however the opposite of "lazy_sweep" appears
to be "immediate_sweep". So use immediate_sweep, and flip
the boolean value of each arg.
I guess this only started failing with r56981 in Dec 2016
("class.c: missing unknown_keyword_error",
commit e3f0cca2f26ba44c810ac980cdff7dda129ae533)
* benchmark/bm_vm1_gc_wb_ary.rb: "lazy_sweep: false" => "immediate_sweep: true"
* benchmark/bm_vm1_gc_wb_ary_promoted.rb: ditto
* benchmark/bm_vm1_gc_wb_obj.rb: ditto
* benchmark/bm_vm1_gc_wb_obj_promoted.rb: ditto
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58565 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Kenta Murata
Takashi Kokubun
Vladimir Dementyev
Ryuta Kamizono
卜部昌平
Nobuyoshi Nakada
S.H
Jeremy Evans
Kazuhiro NISHIYAMA
Lourens Naudé
David Rodríguez
Dylan Thacker-Smith
Masato Ohba
Yaw Boakye
Yusuke Endoh
Aaron Patterson
Samuel Williams
Hiroshi SHIBATA
nobu
eregon
mame
mrkn
shyouhei
k0kubun
tenderlove
samuel
svn
ko1
normal
glass