Bug 213715 – %TypedArray%.prototype.{indexOf,lastIndexOf} are not spec-perfect

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Alexey Shvayka]

%TypedArray%.prototype.{indexOf,lastIndexOf} are not spec-perfect

[Alexey Shvayka]

Created attachment 403037 [details]
Patch

[Alexey Shvayka]

(In reply to Alexey Shvayka from comment #1)
> Created attachment 403037 [details]
> Patch

Warmed-up runs, --outer 48:

                                   r263638                    patch                                       

array-indexOf-empty            221.3595+-2.9548     ^    108.8428+-1.7148        ^ definitely 2.0338x faster
typed-array-indexOf-empty      146.0446+-2.7547     ^    115.9307+-1.8323        ^ definitely 1.2598x faster

<geometric>                    179.6698+-2.0843     ^    112.2352+-1.1371        ^ definitely 1.6008x faster

[Yusuke Suzuki]

Comment on attachment 403037 [details]
Patch

r=me. Can you also ensure that DFG/FTL ArrayIndexOf follows to this?

[Alexey Shvayka]

Created attachment 403530 [details]
Patch

Add DFG/FTL stress test.

[Alexey Shvayka]

(In reply to Yusuke Suzuki from comment #3)
> Comment on attachment 403037 [details]
> Patch
> 
> r=me. Can you also ensure that DFG/FTL ArrayIndexOf follows to this?

ArrayIndexOf OSR exit via SpeculateInt32Operand at SpeculativeJIT::emitPopulateSliceIndex() if `fromIndex` parameter is an object.
That's the reason of such performance boost.
I've added a stress test to ensure ToInteger() after "length" order.

[EWS]

Committed r263944: <https://trac.webkit.org/changeset/263944>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 403530 [details].

[Radar WebKit Bug Importer]

<rdar://problem/65098390>

[Keith Miller]

*** Bug 216817 has been marked as a duplicate of this bug. ***