Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 14400

Summary: optimize jpeg decoder + image renderer
Product: [Eclipse Project] Platform Reporter: Babar Ali <babarali>
Component: SWTAssignee: Christophe Cornu <christophe.cornu+eclipse>
Status: RESOLVED DUPLICATE QA Contact:
Severity: normal    
Priority: P3 CC: carolynmacleod4, mlq.eclipse
Version: 2.0Keywords: helpwanted
Target Milestone: ---   
Hardware: PC   
OS: Windows XP   
Whiteboard:

Description Babar Ali CLA 2002-04-23 08:57:02 EDT 
jpeg too long for decoddecoder needs optimization; it is talking ing big 
images. Moreover imagedata is taking too much memory, + renderer makes the 
whole system lacking in response. 

pictures will be uploaded tonight at 
http://www.picsmart.com/swt-test-images/
Comment 1 Babar Ali CLA 2002-04-23 09:04:04 EDT 
opps something went wrong with my fingers and it was submitted with mangled
text.

here goes again
jpeg decoder is taking too long for big images and requires optimization. 
Moreover imagedata is taking too much memory, + renderer makes the
whole system lacking in response. 

There are two paths for this. existing java code optimization or by using some 
optimized native libraries. I would suggest to go for native as for jpeg and 
tiffs specially when they are big, byte code often happen to lack a lot behind 
native code.
Comment 2 Mike Wilson CLA 2002-04-23 10:27:23 EDT 
Just so you know, we benched GIF image decoding in Java versus the same code 
written as a C native and found the C code to be only about 50% faster (i.e. 
90 seconds for Java code, 60 seconds for C code in the benchmark we were 
running). This is significant, but not enough to solve most of the real-world 
problems.

In any case, I agree that improving our image decoding performance would be a 
good thing. Given the above numbers, I think the problem is more complex than 
simply re-writing the code in C. 

Unfortunately, we aren't going to get to this in the R2.0 time frame. Anyone 
who wants to submit a proposal for improving image processing is welcome to. 
Ideally, this should include scenarios and benchmarks of specific problem 
areas. Even better would be to include a working implementation on all the 
platforms.
Comment 3 Veronika Irvine CLA 2002-09-11 14:06:57 EDT 
Moving from Later.
Comment 4 Christophe Cornu CLA 2005-03-29 10:40:56 EST 

*** This bug has been marked as a duplicate of 24451 ***