Optimize @@species based on a global 'protector' cell
This patch makes ArraySpeciesCreate fast in V8 by avoiding two property reads
when the following conditions are met:
- No Array instance has had its __proto__ reset
- No Array instance has had a constructor property defined
- Array.prototype has not had its constructor changed
- Array[Symbol.species] has not been reset
For subclasses of Array, or for conditions where one of these assumptions is
violated, the full lookup of species is done according to the ArraySpeciesCreate
algorithm. Although this is a "performance cliff", it does not come up in the
expected typical use case of @@species (Array subclassing), so it is hoped that
this can form a good start. Array subclasses will incur the slowness of looking
up @@species, but their use won't slow down invocations of, for example,
Array.prototype.slice on Array base class instances.
Possible future optimizations:
- For the fallback case where the assumptions don't hold, optimize the two
property lookups.
- For Array.prototype.slice and Array.prototype.splice, even if the full lookup
of @@species needs to take place, we still could take the rest of the C++
fastpath. However, to do this correctly requires changing the calling convention
from C++ to JS to pass the @@species out, so it is not attempted in this patch.
With this patch, microbenchmarks of Array.prototype.slice do not suffer a
noticeable performance regression, unlike their previous 2.5x penalty.
TBR=hpayer@chromium.org
Review URL: https://codereview.chromium.org/1689733002
Cr-Commit-Position: refs/heads/master@{#34199}
2016-02-22 21:01:29 +00:00
|
|
|
// Copyright 2016 the V8 project authors. All rights reserved.
|
|
|
|
// Use of this source code is governed by a BSD-style license that can be
|
|
|
|
// found in the LICENSE file.
|
|
|
|
|
Make array __proto__ manipulations not disturb the species protector
Previously, the species protector was invalidated whenever the __proto__ of
an Array instance was manipulated. Then, if the map's new_target_is_base field
remained set, it was correct to conclude that GetPrototypeOf(array) was
%ArrayPrototype%. However, this choice caused the popular D3 framework to
invalidate the species protector, causing many functions to become slower.
This patch eliminates that aspect of the species protector. Instead, the check
is to look at the instance->map()->prototype(). It is valid to look directly
at the map's prototype slot, ignoring hidden prototypes and proxies, because
- This is only called on Array instances, so the receiver cannot be a Proxy.
- For hidden prototypes, any inaccuracy would only result in conservatively
taking the slow path.
Theoretically, this patch could make methods applied to arrays from other
contexts slower. However, the slowdown would only affect a particular array
instance and not have a global spill-over effect. Further, the slowdown could
be addressed by tracking, either in the instance's map or in the actual
prototype object, whether it is a %ArrayPrototype% from any context, in a way
which is cheap to query, and use that rather than comparing to the currently
executing native context.
In interactive testing, this patch led the OnShape CAD system to experience
faster load times (110+s -> 40s).
BUG=chromium:606207
LOG=Y
Review-Url: https://codereview.chromium.org/1936393002
Cr-Commit-Position: refs/heads/master@{#36033}
2016-05-04 16:47:34 +00:00
|
|
|
// Flags: --harmony-species --allow-natives-syntax
|
Optimize @@species based on a global 'protector' cell
This patch makes ArraySpeciesCreate fast in V8 by avoiding two property reads
when the following conditions are met:
- No Array instance has had its __proto__ reset
- No Array instance has had a constructor property defined
- Array.prototype has not had its constructor changed
- Array[Symbol.species] has not been reset
For subclasses of Array, or for conditions where one of these assumptions is
violated, the full lookup of species is done according to the ArraySpeciesCreate
algorithm. Although this is a "performance cliff", it does not come up in the
expected typical use case of @@species (Array subclassing), so it is hoped that
this can form a good start. Array subclasses will incur the slowness of looking
up @@species, but their use won't slow down invocations of, for example,
Array.prototype.slice on Array base class instances.
Possible future optimizations:
- For the fallback case where the assumptions don't hold, optimize the two
property lookups.
- For Array.prototype.slice and Array.prototype.splice, even if the full lookup
of @@species needs to take place, we still could take the rest of the C++
fastpath. However, to do this correctly requires changing the calling convention
from C++ to JS to pass the @@species out, so it is not attempted in this patch.
With this patch, microbenchmarks of Array.prototype.slice do not suffer a
noticeable performance regression, unlike their previous 2.5x penalty.
TBR=hpayer@chromium.org
Review URL: https://codereview.chromium.org/1689733002
Cr-Commit-Position: refs/heads/master@{#34199}
2016-02-22 21:01:29 +00:00
|
|
|
|
|
|
|
// Overwriting Array[Symbol.species] updates the protector
|
|
|
|
|
|
|
|
let x = [];
|
|
|
|
|
|
|
|
assertEquals(Array, x.map(()=>{}).constructor);
|
|
|
|
assertEquals(Array, x.filter(()=>{}).constructor);
|
|
|
|
assertEquals(Array, x.slice().constructor);
|
|
|
|
assertEquals(Array, x.splice().constructor);
|
|
|
|
assertEquals(Array, x.concat([1]).constructor);
|
|
|
|
assertEquals(1, x.concat([1])[0]);
|
|
|
|
|
|
|
|
class MyArray extends Array { }
|
|
|
|
|
|
|
|
Object.defineProperty(Array, Symbol.species, {value: MyArray});
|
Make array __proto__ manipulations not disturb the species protector
Previously, the species protector was invalidated whenever the __proto__ of
an Array instance was manipulated. Then, if the map's new_target_is_base field
remained set, it was correct to conclude that GetPrototypeOf(array) was
%ArrayPrototype%. However, this choice caused the popular D3 framework to
invalidate the species protector, causing many functions to become slower.
This patch eliminates that aspect of the species protector. Instead, the check
is to look at the instance->map()->prototype(). It is valid to look directly
at the map's prototype slot, ignoring hidden prototypes and proxies, because
- This is only called on Array instances, so the receiver cannot be a Proxy.
- For hidden prototypes, any inaccuracy would only result in conservatively
taking the slow path.
Theoretically, this patch could make methods applied to arrays from other
contexts slower. However, the slowdown would only affect a particular array
instance and not have a global spill-over effect. Further, the slowdown could
be addressed by tracking, either in the instance's map or in the actual
prototype object, whether it is a %ArrayPrototype% from any context, in a way
which is cheap to query, and use that rather than comparing to the currently
executing native context.
In interactive testing, this patch led the OnShape CAD system to experience
faster load times (110+s -> 40s).
BUG=chromium:606207
LOG=Y
Review-Url: https://codereview.chromium.org/1936393002
Cr-Commit-Position: refs/heads/master@{#36033}
2016-05-04 16:47:34 +00:00
|
|
|
assertFalse(%SpeciesProtector());
|
Optimize @@species based on a global 'protector' cell
This patch makes ArraySpeciesCreate fast in V8 by avoiding two property reads
when the following conditions are met:
- No Array instance has had its __proto__ reset
- No Array instance has had a constructor property defined
- Array.prototype has not had its constructor changed
- Array[Symbol.species] has not been reset
For subclasses of Array, or for conditions where one of these assumptions is
violated, the full lookup of species is done according to the ArraySpeciesCreate
algorithm. Although this is a "performance cliff", it does not come up in the
expected typical use case of @@species (Array subclassing), so it is hoped that
this can form a good start. Array subclasses will incur the slowness of looking
up @@species, but their use won't slow down invocations of, for example,
Array.prototype.slice on Array base class instances.
Possible future optimizations:
- For the fallback case where the assumptions don't hold, optimize the two
property lookups.
- For Array.prototype.slice and Array.prototype.splice, even if the full lookup
of @@species needs to take place, we still could take the rest of the C++
fastpath. However, to do this correctly requires changing the calling convention
from C++ to JS to pass the @@species out, so it is not attempted in this patch.
With this patch, microbenchmarks of Array.prototype.slice do not suffer a
noticeable performance regression, unlike their previous 2.5x penalty.
TBR=hpayer@chromium.org
Review URL: https://codereview.chromium.org/1689733002
Cr-Commit-Position: refs/heads/master@{#34199}
2016-02-22 21:01:29 +00:00
|
|
|
|
|
|
|
assertEquals(MyArray, x.map(()=>{}).constructor);
|
|
|
|
assertEquals(MyArray, x.filter(()=>{}).constructor);
|
|
|
|
assertEquals(MyArray, x.slice().constructor);
|
|
|
|
assertEquals(MyArray, x.splice().constructor);
|
|
|
|
assertEquals(MyArray, x.concat([1]).constructor);
|
|
|
|
assertEquals(1, x.concat([1])[0]);
|