Use a BOP for Hash#default #6945
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
This does seem to improve Railsbench a little bit. There are no real changes to YJIT here (just bindgen updates) but it should improve its performance by the same as the interpreter because it also just calls Before After |
nitpick: It seems more easier to understand the difference this PR is making if you compare before/after between interpreters and YJITs, not interp/yjit. You could do it like: If you intentionally do it that way to avoid having multiple build directories, it's okay too :) But I just wanted to make sure you're aware. |
|
I should do that 😅. When I'm testing just for myself I tend to run the comparison Ruby directly from the source tree, but that has also caused me to make mistakes. |
0b0ee01 to
ff0336a
Compare
On a hash miss we need to call default if it is redefined in order to
return the default value to be used. Previously we checked this with
rb_method_basic_definition_p, which avoids the method call but requires
a method lookup.
This commit replaces the previous check with BASIC_OP_UNREDEFINED_P and
a new BOP_DEFAULT. We still need to fall back to
rb_method_basic_definition_p when called on a subclasss of hash.
| |compare-ruby|built-ruby|
|:---------------|-----------:|---------:|
|hash_aref_miss | 2.692| 3.531|
| | -| 1.31x|
Co-authored-by: Daniel Colson <danieljamescolson@gmail.com>
Co-authored-by: "Ian C. Anderson" <ian@iancanderson.com>
Co-authored-by: Jack McCracken <me@jackmc.xyz>
We should always have a T_HASH here, so we can use FL_TEST_RAW to avoid checking whether we may have an immediate value. I expect this to be a very small performance improvement (perf stat ./miniruby benchmark/hash_aref_miss.rb shows a ~1% improvement). It also removes 9 instructions from rb_hash_default_value on x86_64.
ff0336a to
bf6099b
Compare
matzbot
pushed a commit
that referenced
this pull request
Nov 8, 2023
Since #6945 the extension dir changed to Gem::BasicSpecification's implementation, we didn't hook that in rubygems_ext.rb. So for universal rubies, we ended up using the universal platform name when installing, but arch replaced platform name when checking. This lead to native extensions can never be correctly installed on universal rubies. Hook Gem::BasicSpecifications so the behavior is consistent on installing and checking. rubygems/rubygems@8d699ed096
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
On a hash miss we need to call default if it is redefined in order to return the default value to be used. Previously we checked this with rb_method_basic_definition_p, which avoids the method call but requires a method lookup.
This commit replaces the previous check with BASIC_OP_UNREDEFINED_P and a new BOP_DEFAULT. We still need to fall back to rb_method_basic_definition_p when called on a subclass of hash.
On railsbench, before this change I measured how many times we call
rb_method_basic_definition_pand with what IDs. For 5 iterations:before: 5086076
after: 716854
So this is optimizing ~95% of the checks. Rails does include some subclasses of Hash (ex. HashWithIndifferentAccess), which is why it isn't 100%.
defaultwas by far the ID most passed to rb_method_basic_definition_p, second wasto_procwith 273898 calls.