Webkit aka Safari Version 3.1.2 (5525.20.1) vs Minefield aka Firefox vers 3.1b2

Update: The current Webkit nightly r37894 is even faster than Minefield when it comes to the SunSpider Javascript benchmark.

TEST                   COMPARISON            FROM                 TO             DETAILS

=============================================================================

** TOTAL **:           2.03x as fast     1511.6ms +/- 3.7%   745.0ms +/- 0.4%     significant

=============================================================================

  3d:                  1.07x as fast      142.4ms +/- 2.5%   133.4ms +/- 0.8%     significant
    cube:              1.27x as fast       62.6ms +/- 3.6%    49.4ms +/- 3.8%     significant
    morph:             *1.55x as slow*     33.2ms +/- 4.9%    51.6ms +/- 1.3%     significant
    raytrace:          1.44x as fast       46.6ms +/- 1.5%    32.4ms +/- 2.1%     significant

  access:              2.66x as fast      182.2ms +/- 18.6%    68.4ms +/- 2.1%     significant
    binary-trees:      5.17x as fast       47.6ms +/- 1.4%     9.2ms +/- 6.0%     significant
    fannkuch:          5.68x as fast       89.8ms +/- 38.1%    15.8ms +/- 3.5%     significant
    nbody:             *1.15x as slow*     31.2ms +/- 1.8%    36.0ms +/- 2.4%     significant
    nsieve:            1.84x as fast       13.6ms +/- 5.0%     7.4ms +/- 9.2%     significant

  bitops:              1.36x as fast       53.2ms +/- 24.0%    39.2ms +/- 2.7%     significant
    3bit-bits-in-byte: *3.00x as slow*      1.6ms +/- 42.6%     4.8ms +/- 11.6%     significant
    bits-in-byte:      -                    9.8ms +/- 44.3%     7.2ms +/- 7.7% 
    bitwise-and:       -                   13.4ms +/- 53.9%     8.8ms +/- 6.3% 
    nsieve-bits:       1.54x as fast       28.4ms +/- 16.6%    18.4ms +/- 3.7%     significant

  controlflow:         8.71x as fast       36.6ms +/- 1.9%     4.2ms +/- 13.2%     significant
    recursive:         8.71x as fast       36.6ms +/- 1.9%     4.2ms +/- 13.2%     significant

  crypto:              1.50x as fast       65.0ms +/- 10.7%    43.4ms +/- 1.6%     significant
    aes:               2.54x as fast       34.6ms +/- 2.0%    13.6ms +/- 5.0%     significant
    md5:               1.53x as fast       23.0ms +/- 27.3%    15.0ms +/- 0.0%     significant
    sha1:              *2.00x as slow*      7.4ms +/- 9.2%    14.8ms +/- 3.8%     significant

  date:                3.55x as fast      265.6ms +/- 0.4%    74.8ms +/- 0.7%     significant
    format-tofte:      4.73x as fast      148.4ms +/- 0.7%    31.4ms +/- 2.2%     significant
    format-xparb:      2.70x as fast      117.2ms +/- 1.4%    43.4ms +/- 1.6%     significant

  math:                *2.02x as slow*     50.0ms +/- 5.0%   100.8ms +/- 0.6%     significant
    cordic:            *1.45x as slow*     24.2ms +/- 9.2%    35.0ms +/- 0.0%     significant
    partial-sums:      *2.91x as slow*     17.2ms +/- 6.0%    50.0ms +/- 0.0%     significant
    spectral-norm:     *1.84x as slow*      8.6ms +/- 7.9%    15.8ms +/- 3.5%     significant

  regexp:              6.15x as fast      237.4ms +/- 0.6%    38.6ms +/- 1.8%     significant
    dna:               6.15x as fast      237.4ms +/- 0.6%    38.6ms +/- 1.8%     significant

  string:              1.98x as fast      479.2ms +/- 6.5%   242.2ms +/- 0.7%     significant
    base64:            1.04x as fast       20.6ms +/- 3.3%    19.8ms +/- 2.8%     significant
    fasta:             3.91x as fast      134.6ms +/- 22.6%    34.4ms +/- 2.0%     significant
    tagcloud:          1.30x as fast      112.8ms +/- 0.9%    86.6ms +/- 2.4%     significant
    unpack-code:       2.90x as fast      160.0ms +/- 0.5%    55.2ms +/- 1.0%     significant
    validate-input:    1.11x as fast       51.2ms +/- 2.7%    46.2ms +/- 4.4%     significant

They weren't joking when they said Javascript performance has been improved. Safari Version 3.1.2 (5525.20.1) as a comparison.

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total:                 3082.4ms +/- 0.2%
--------------------------------------------

  3d:                   375.2ms +/- 0.4%
    cube:               122.2ms +/- 1.1%
    morph:              129.8ms +/- 0.8%
    raytrace:           123.2ms +/- 0.5%

  access:               486.4ms +/- 0.6%
    binary-trees:        63.8ms +/- 1.6%
    fannkuch:           230.8ms +/- 1.0%
    nbody:              135.4ms +/- 0.5%
    nsieve:              56.4ms +/- 2.5%

  bitops:               425.4ms +/- 0.7%
    3bit-bits-in-byte:   62.6ms +/- 1.1%
    bits-in-byte:        94.4ms +/- 1.2%
    bitwise-and:        164.2ms +/- 1.6%
    nsieve-bits:        104.2ms +/- 0.5%

  controlflow:           82.6ms +/- 0.8%
    recursive:           82.6ms +/- 0.8%

  crypto:               221.6ms +/- 0.5%
    aes:                 73.8ms +/- 0.8%
    md5:                 74.0ms +/- 0.0%
    sha1:                73.8ms +/- 0.8%

  date:                 265.8ms +/- 0.5%
    format-tofte:       122.6ms +/- 0.9%
    format-xparb:       143.2ms +/- 0.4%

  math:                 417.4ms +/- 0.6%
    cordic:             161.2ms +/- 0.6%
    partial-sums:       179.6ms +/- 0.9%
    spectral-norm:       76.6ms +/- 1.5%

  regexp:               199.6ms +/- 0.3%
    dna:                199.6ms +/- 0.3%

  string:               608.4ms +/- 0.5%
    base64:              89.2ms +/- 1.5%
    fasta:              167.4ms +/- 0.8%
    tagcloud:           122.4ms +/- 0.6%
    unpack-code:        123.8ms +/- 0.4%
    validate-input:     105.6ms +/- 1.1%

My setup: iMac 2.4 GHz/4 GB RAM/OS X 10.5.5

Download the latest Webkit nightly builds here.

Download the latest Minefield nightly builds here.

So it seems the Webkit nightly is now about 4x faster than Safari. :D

Let the gloating begin

So Apple recently announced their quarterly results and the iPhone is an unprecedented success, firmly entrenched as a viable portable computing device. John Gruber at Daring Fireball and various other Mac-oriented websites are roasting the "analysts" and "experts" who got it so wrong.

Here are some quick thoughts:

1. The App Store. This is the real killer app. Building on Apple's vast experience in online music delivery, installing software couldn't be simpler. Until a competitor understands this and offers a much better alternative, Apple will own this market. It's only a matter of time.

2. Android will not gain mainstream market acceptance. Android is great for tinkerers or people who demand total control of the phone. But since each handset maker can implement their own interface and applications, there is no coherent vision to tie everything together. This makes it hard to market the product well, and also leads to many technical and support issues. The only way for these licensors to differentiate themselves would be based on price, and we know where that leads to.

3. Every engineer should have this Einstein quote on their minds: "Everything should be made as simple as possible, but not simpler." Adding features should not introduce undue complexity, especially when proposed feature is not essential, used infrequently and/or yields insignificant value.

4. Blindly copying features without understanding the context will lead to grief. HP introduced the TouchSmart PC to great fanfare, it is not the first attempt at a touch-screen PC and still has nothing to show for it. Why? HP does not offer any compelling application that demands the use of a touchscreen. Also, it is terrible from a usability and productivity standpoint; I've got to get my hands off the keyboard and mess with the screen just to do something which can already be accomplished with a mouse and keyboard. Put another way, HP chickened out and did not have the courage to completely remove the keyboard and mouse. Not that I blame them, having your hands over a display for prolonged periods of time would be tiring. A multitouch interface on a small portable device on the other hand makes perfect sense because you are constrained by size, weight and ergonomics decisions