Japanese text rendering by 2D with JFC compoenent is considerably very slow.
Try open file kanji.* attached using Notepad under JFC demo in the release. Size of file is same but content is different.
kanji.sjis containes all sjis char but kanji.ascii only containes ASCII character.
CPU time usage(user time) to complete open and display a file taken by "time" of MKS toolkit:
JDK1.2beta3n JDK1.1.5
kanji.ascii 8.86s 0.01s
kanji.sjis 18.98s 0.01s
Measurement was taken on
WinNT version 4.0 Japanese running on P2 300MHz w/ 128M RAM.
Acceptable degradation of speed performance to render Japanese character
is 20%. So, it should be arround 9sec.
And it is really painfull to scroll up and down the text if Japanese.
(It looks not scrolling but paging.)
[jim.hu@prc 1998-07-10]
This is also a big problem for Chinese. I have attached a program
and data files to reproduce it. You can run it in C, ja and zh
locale to see its performance.
uncompress stresstester.tar.Z
tar xvf stresstester.tar
type runit
press the button "Load text".
Try open file kanji.* attached using Notepad under JFC demo in the release. Size of file is same but content is different.
kanji.sjis containes all sjis char but kanji.ascii only containes ASCII character.
CPU time usage(user time) to complete open and display a file taken by "time" of MKS toolkit:
JDK1.2beta3n JDK1.1.5
kanji.ascii 8.86s 0.01s
kanji.sjis 18.98s 0.01s
Measurement was taken on
WinNT version 4.0 Japanese running on P2 300MHz w/ 128M RAM.
Acceptable degradation of speed performance to render Japanese character
is 20%. So, it should be arround 9sec.
And it is really painfull to scroll up and down the text if Japanese.
(It looks not scrolling but paging.)
[jim.hu@prc 1998-07-10]
This is also a big problem for Chinese. I have attached a program
and data files to reproduce it. You can run it in C, ja and zh
locale to see its performance.
uncompress stresstester.tar.Z
tar xvf stresstester.tar
type runit
press the button "Load text".