Updates:
Status: Accepted
Owner: [email protected]
Comment #2 on issue 957 by [email protected]: bad V8 performance on
floating point intensive cases
http://code.google.com/p/v8/issues/detail?id=957
Our latest version 3.0.0 (available as of today) improves many floating
point-intensive benchmarks significantly. We added the possibility to do FP
computations unboxed on local doubles.
You should see around 2x on imaging-darkroom and imaging-desaturate on
V8-ia32 compared to previous versions of V8.
We will continue to improve our performance on arithmetic operations even
further and bring the same performance to our other platforms as well.
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
