EclipseZone - Eclipse 3.2 M6 JDT Performance ...

The Rich Engineering Heritage Behind Dependency Injection

Andrew McVeigh takes us on a tour of the rich heritage behind dependency injection, what it represents, and tells us why its here to stay.

Java, the OLPC, and community responsibility

The "One Laptop Per Child" project has a great device ready to ship, but there's no Java on there. Let's think about working together to put Java on OLPC!

Replies: 7 - Pages: 1

Reply 1. At 2:35 AM on Apr 10, 2006, Jilles van Gurp wrote:: Re: Eclipse 3.2 M6 JDT Performance
I wouldn't call the improvement over M5 dramatic but its definately there. Since eclipse 3.0, the instances where eclipse seems to hang for several seconds has been decreasing. I had very little complaints about M5 already but I did encounter a few of those hangs (e.g. autocompletion) which no longer seem to be there in M6. So that's good.

Overall, the eclipse developers seem to be doing a great job improving features and performance. For most low end hardware, 3.2M6 is probably a better eclipse than any of the previous 3.x versions. It uses less memory, wastes less cpu cycles and packs more features. I think the approach of measuring performance and addressing the bottlenecks is a really great way to develop for quality.
Reply 2. At 6:02 AM on Apr 10, 2006, murphee (Werner Schuster) wrote:: Re: Eclipse 3.2 M6 JDT Performance
> its definately there. Since eclipse 3.0, the
> instances where eclipse seems to hang for several
> seconds has been decreasing.

About these hangs for several seconds: have you tried increasing the Xmx setting of your Eclipse installation? I had these problems too (mostly after a long day with a large workspace), and the hangs were due to a nearly-but-not-quite full memory, which caused full GC runs to happen more often (full GC runs pause the whole JVM), instead of the incremental GC.

After I increased the Xmx the behavior vanished. I suppose, if the memory usage decreased in recent milestones, this might have improved this situation, because the memory limit is reached much later (or not at all).
Reply 3. At 8:19 AM on Apr 10, 2006, Philippe Mulet wrote:: Re: Eclipse 3.2 M6 JDT Performance
Thanks for noticing. There were many changes related to make the live editing experience smoother. We made the model caching policy better scalable, by making it more fine grain; we also better reuse allocated objects. The compiler now is sorting fields & methods for allowing binary searching, etc...

In the end, we also noticed that our memory requirement is significantly lower than in the past. This still needs to be proven, but early experimentation tells us that when -Xmx256M was required, only -Xmx128M seems now necessary to self-host Eclipse development.

There are a few more performance tuning in progress, so if you did like M6, you should love RC1.
Reply 4. At 10:42 AM on Apr 10, 2006, Fred Grott wrote:: A performance question
the performance inceases fro JDT are nice. But I have a related question

Has anyone notice a CPU 100% useage when the Help view is open and doing task such as a plugin/feature update on winXP?

If I close the Help View it goes back to what I am use to seeing as far as CPU % of use.

Happens when I have the full help contents opne in help view.; Fred Grott, ShareMe jabber:fg82183@conference.javalobby.org(shareme) Official contributor to EclipseZone.com and supporter of JavaLobby.org.
Reply 5. At 7:52 PM on Apr 12, 2006, Romano Silva wrote:: Re: Eclipse 3.2 M6 JDT Performance
I would agree that I have seen some performance improvements in compiling, but simple things that I used to do in 3.2M4 are now making me nuts. I can say for certain that my productivity decreased this week since I installed 3.2M6. I am rolling back to 3.2M4 to verify these issues:

1- CTRL+O to open outline cause constant hang
2- CTRL+SHIFT-T to open type list cause constant hang
3- F3 to open reference cause constant hang
Reply 6. At 6:09 AM on Apr 13, 2006, Philippe Mulet wrote:: Re: Eclipse 3.2 M6 JDT Performance
Please enter bug reports asap with steps to reproduce.
We aren't aware of such issues.
Reply 7. At 6:49 AM on Apr 13, 2006, Philippe Mulet wrote:: Re: Eclipse 3.2 M6 JDT Performance
I wonder if you are seeing:
http://bugs.eclipse.org/bugs/show_bug.cgi?id=134172

If this is the case, it already is fixed in RC1. Can you pls double check this ?