Javascript speed in the webbrowser Opera 10 got reduced by a factor 3! Or has it? Cause and solution.
A.E.Veltstra
Saturday, April 25, 2009
The oddest thing just occurred. I tested the javascript speed in Opera 10 (build 1426), using my favourite javascript application (the YASS described in an earlier post), and a very crude way of benchmarking (counting seconds while waiting for the application to finish solving the given Sudoku puzzle).
With my Opera 9.63 configuration, my javascript application takes roughly 8 seconds to solve the puzzle. In comparison: MS Internet Explorer 8 requires roughly 15 seconds.
To my amazement, the new Opera 10 required even more than that: nearly 24 seconds.
Now, for reasons only known to me, I performed that test before I switched Opera's Appearance Skin from Opera Standard to Windows Native (yes, I run a MS Windows XP Home edition on a rather fast desktop computer). After switching, something compelled me to run the test again... just to make sure I hadn't miscounted.
Lo and behold: it took the application 8 seconds again.
Huh? So wait... the choice of the Opera Appearance Skin influences the speed of its javascript engine?
Apparently.
This might only be the case, due to my application echoing intermediate results back to the screen. However, that doesn't strike me as particularly far-fetched.
Yes, it can probably run faster if the echo is done on interval, but oh my, showing the Opera Standard Skin slows down my javascript application by a factor of nearly 3...
I know some competitors who are going to love sneering at this...
Edited to add: this problem seems either solved or reduced to imperception using Opera 10.5 pre-alpha.
(The ΩJr.Net is not affiliated with Opera SA nor Microsoft Inc.)
With my Opera 9.63 configuration, my javascript application takes roughly 8 seconds to solve the puzzle. In comparison: MS Internet Explorer 8 requires roughly 15 seconds.
To my amazement, the new Opera 10 required even more than that: nearly 24 seconds.
Now, for reasons only known to me, I performed that test before I switched Opera's Appearance Skin from Opera Standard to Windows Native (yes, I run a MS Windows XP Home edition on a rather fast desktop computer). After switching, something compelled me to run the test again... just to make sure I hadn't miscounted.
Lo and behold: it took the application 8 seconds again.
Huh? So wait... the choice of the Opera Appearance Skin influences the speed of its javascript engine?
Apparently.
This might only be the case, due to my application echoing intermediate results back to the screen. However, that doesn't strike me as particularly far-fetched.
Yes, it can probably run faster if the echo is done on interval, but oh my, showing the Opera Standard Skin slows down my javascript application by a factor of nearly 3...
I know some competitors who are going to love sneering at this...
Edited to add: this problem seems either solved or reduced to imperception using Opera 10.5 pre-alpha.
(The ΩJr.Net is not affiliated with Opera SA nor Microsoft Inc.)
Need problem solving?
Talk to me. Let's meet for coffee or over lunch. Mail me at “omegajunior at protonmail dot com”.