[Webkit-unassigned] [Bug 98586] Extend random precision to 53bits

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=98586





--- Comment #10 from Yusuke Suzuki <utatane.tea at gmail.com>  2012-10-07 02:00:22 PST ---
(In reply to comment #2)
> (From update of attachment 167443 [details])
> What's the performance impact of this?

I've run SunSpider.

Before applying patch
============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total:                 143.6ms +/- 2.7%
--------------------------------------------

  3d:                   25.5ms +/- 5.3%
    cube:                8.5ms +/- 4.4%
    morph:               6.2ms +/- 7.3%
    raytrace:           10.8ms +/- 6.8%

  access:               11.4ms +/- 4.4%
    binary-trees:        1.0ms +/- 0.0%
    fannkuch:            5.1ms +/- 4.4%
    nbody:               3.0ms +/- 0.0%
    nsieve:              2.3ms +/- 15.0%

  bitops:                7.1ms +/- 3.2%
    3bit-bits-in-byte:   1.0ms +/- 0.0%
    bits-in-byte:        2.0ms +/- 0.0%
    bitwise-and:         1.1ms +/- 20.5%
    nsieve-bits:         3.0ms +/- 0.0%

  controlflow:           1.2ms +/- 25.1%
    recursive:           1.2ms +/- 25.1%

  crypto:               12.7ms +/- 2.7%
    aes:                 7.7ms +/- 4.5%
    md5:                 3.0ms +/- 0.0%
    sha1:                2.0ms +/- 0.0%

  date:                 19.6ms +/- 2.5%
    format-tofte:       10.7ms +/- 3.2%
    format-xparb:        8.9ms +/- 2.5%

  math:                 11.7ms +/- 5.8%
    cordic:              3.0ms +/- 0.0%
    partial-sums:        6.7ms +/- 10.1%
    spectral-norm:       2.0ms +/- 0.0%

  regexp:                9.0ms +/- 9.2%
    dna:                 9.0ms +/- 9.2%

  string:               45.4ms +/- 3.3%
    base64:              4.0ms +/- 0.0%
    fasta:               5.8ms +/- 5.2%
    tagcloud:           10.9ms +/- 7.8%
    unpack-code:        19.6ms +/- 3.9%
    validate-input:      5.1ms +/- 4.4%

After applying patch
============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total:                 144.1ms +/- 1.7%
--------------------------------------------

  3d:                   26.7ms +/- 6.6%
    cube:                9.1ms +/- 5.8%
    morph:               6.4ms +/- 9.4%
    raytrace:           11.2ms +/- 7.2%

  access:               11.3ms +/- 4.3%
    binary-trees:        1.1ms +/- 20.5%
    fannkuch:            5.0ms +/- 0.0%
    nbody:               3.0ms +/- 0.0%
    nsieve:              2.2ms +/- 13.7%

  bitops:                7.4ms +/- 5.0%
    3bit-bits-in-byte:   1.0ms +/- 0.0%
    bits-in-byte:        2.0ms +/- 0.0%
    bitwise-and:         1.4ms +/- 26.4%
    nsieve-bits:         3.0ms +/- 0.0%

  controlflow:           1.2ms +/- 25.1%
    recursive:           1.2ms +/- 25.1%

  crypto:               12.9ms +/- 1.8%
    aes:                 7.9ms +/- 2.9%
    md5:                 3.0ms +/- 0.0%
    sha1:                2.0ms +/- 0.0%

  date:                 19.8ms +/- 1.5%
    format-tofte:       10.8ms +/- 2.8%
    format-xparb:        9.0ms +/- 0.0%

  math:                 11.0ms +/- 0.0%
    cordic:              3.0ms +/- 0.0%
    partial-sums:        6.0ms +/- 0.0%
    spectral-norm:       2.0ms +/- 0.0%

  regexp:                8.9ms +/- 2.5%
    dna:                 8.9ms +/- 2.5%

  string:               44.9ms +/- 2.2%
    base64:              4.0ms +/- 0.0%
    fasta:               5.8ms +/- 5.2%
    tagcloud:           10.6ms +/- 3.5%
    unpack-code:        19.4ms +/- 2.6%
    validate-input:      5.1ms +/- 4.4%

I think the performance impact of this is small and acceptable.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.