According to Masaki Takase, the PS3's JavaScript performance beats that of IE7's.
Really? That sounds like something that's testable! Cue SunSpider!
Well, actually, no. Because the PS3 browser can't handle all the SunSpider tests. I had to remove three of them to get it to complete. Results in the full article.
To compare, I ran IE7 under Windows XP on an Intel Core Duo T2600 (2.16GHz) - a PC that was bought before the PS3 was actually released. So this should be a relatively "fair" comparison, as both pieces of hardware date to about the same time.
The version of IE7 is fully patched. Keep in mind that the total score for the IE7 build is biased against it, because it includes tests the PS3 skipped, raising its total time by over 12 seconds spent running tests the PS3 browser simply couldn't handle.
| PS3 | IE7 | |||
|---|---|---|---|---|
| Test | Avg. Time | Conf. Int | Avg. Time | Conf. Int |
| Total | 98001.0ms | ±0.1% | 35491.8ms | ±2.0% |
| 3d | 10698.6ms | ±0.5% | 1500.4ms | ±4.7% |
| cube | 2482.4ms | ±0.7% | 425.2ms | ±13.9% |
| morph | 5386.0ms | ±0.8% | 500.0ms | ±9.1% |
| raytrace | 2830.2ms | ±0.7% | 575.2ms | ±7.3% |
| access | 13422.0ms | ±0.9% | 2337.0ms | ±3.0% |
| binary-trees | 1851.0ms | ±1.0% | 524.8ms | ±3.3% |
| fannkuch | 7822.6ms | ±1.0% | 915.6ms | ±6.1% |
| nbody | 1662.2ms | ±1.9% | 431.0ms | ±11.0% |
| nsieve | 2086.2ms | ±2.3% | 465.6ms | ±1.9% |
| bitops | 16259.0ms | ±0.4% | 2022.8ms | ±2.6% |
| 3bit-bits-in-byte | 4377.2ms | ±0.4% | 472.0ms | ±10.3% |
| bits-in-byte | 5297.4ms | ±0.5% | 475.2ms | ±11.0% |
| bitwise-and | Failed (JavaScript error) | 578.4ms | ±7.5% | |
| nsieve-bits | 6584.4ms | ±0.7% | 497.2ms | ±8.5% |
| controlflow | No tests completed | 547.0ms | ±9.1% | |
| recursive | CRASHED | 547.0ms | ±9.1% | |
| crypto | 10504.6ms | ±1.2% | 1259.2ms | ±3.6% |
| aes | 2996.6ms | ±0.5% | 475.2ms | ±4.7% |
| md5 | 4029.4ms | ±3.4% | 387.4ms | ±11.4% |
| sha1 | 3478.6ms | ±0.3% | 396.6ms | ±4.3% |
| date | 5178.2ms | ±0.8% | 1531.4ms | ±2.9% |
| format-tofte | 2091.6ms | ±0.6% | 1044.0ms | ±3.4% |
| format-xparb | 3086.6ms | ±1.1% | 487.4ms | ±7.6% |
| math | 17374.4ms | ±0.5% | 1362.4ms | ±4.7% |
| cordic | 5594.0ms | ±0.4% | 559.4ms | ±9.0% |
| partial-sums | 9235.6ms | ±0.7% | 365.6ms | ±14.4% |
| spectral-norm | 2544.8ms | ±1.2% | 437.4ms | ±10.4% |
| regexp | 3083.8ms | ±1.4% | 481.2ms | ±10.1% |
| dna | 3083.8ms | ±1.4% | 481.2ms | ±10.1% |
| string | 21480.4ms | ±0.5% | 24450.4ms | ±3.3% |
| base64 | Failed (JavaScript error) | 11528.2ms | ±6.1% | |
| fasta | 6818.4ms | ±1.3% | 550.2ms | ±7.3% |
| tagcloud | 4058.8ms | ±1.1% | 3481.4ms | ±2.1% |
| unpack-code | 7951.6ms | ±0.6% | 534.2ms | ±3.1% |
| validate-input | 2651.6ms | ±2.0% | 8356.4ms | ±1.1% |
Oh yeah. Clearly a better performer than IE7.
Some more results for comparison
I thought that I'd add some more results just for comparison purposes. All of these tests were done on an even older computer: an AMD Athlon 64 3400+ running openSUSE 11.0 or 11.1 beta3. Here are my results:
===========================================================
RESULTS (means and 95% confidence intervals)
-----------------------------------------------------------
Opera 9.61, openSUSE 11.0: 9947.2ms +/- 11.9%
-----------------------------------------------------------
Konqueror 3.5.9, openSUSE 11.0: 42003.8ms +/- 2.6%
-----------------------------------------------------------
Konqueror 4.1.2, openSUSE 11.1 beta3: 12253.6ms +/- 1.3%
-----------------------------------------------------------
Firefox 3.0.3, openSUSE 11.0: 5542.6ms +/- 13.4%
-----------------------------------------------------------
Firefox 3.0.3, openSUSE 11.1 beta3: 9294.6ms +/- 4.5%
-----------------------------------------------------------
Notes:
Opera not yet installed on openSUSE 11.1 beta3
string-validate-input caused warning about KHTML freeze on Konqueror 3.5.9. Manual continuation was required on each pass for the test to complete.
Even in the worst case (where I had to manually click a button to get the test to continue) it still more than twice as fast as a PS/3.
That doesn't really mean anything, though
Sony never claimed that their JavaScript performance was better than Opera or Firefox, and explicitly said it wasn't up to WebKit standards.
On the other hand, both the Wii and the iPod Touch (!!) have better results on the benchmark than the PS3 browser! And both of them were able to complete all benchmarks without crashing!
I'll be posting full results eventually.
It means something..
It means the PS3, for all of its processing power, doesn't do sell well with JavaScript against any other browser. Also, even against an older computer, IE7 is far slower by comparison to the competition. Then again, you are dealing with Vista.