Bug #4360
open
Kepler gets slow under XP
Added by Oliver Soong about 15 years ago.
Updated almost 15 years ago.
Description
I've been holding off on this, because it's hard to demonstrate, OS-specific, and quite difficult to debug, but it's getting to be problematic. Kepler slows significantly under XP. It's noticeable following the example of http://bugzilla.ecoinformatics.org/show_bug.cgi?id=4359. This particular bug may exacerbate the problem, but it has been noticeable in other unrelated situations. I'd quote duration statistics, but much of the time is spent in the Preinitializing and Wrapping Up which aren't included in the duration.
I'd argue for the importance of this for the release. My understanding is that most devs use macs, while much of our user-base (at least in South Africa, but likely elsewhere as well) uses XP.
It's possible this is related to the Java VM version, but I'm not certain about that.
Here at Berkeley, lots if people use Ptolemy under XP. I run Ptolemy regularly under Windows 2003 Server R2.
The most likely cause of slowness would be running out of memory. Memory is set by invoking Kepler with the -Xmx option. I believe Kepler is being
invoked with -Xmx512M, which means a maximum of 512 megabytes are allocated.
Increasing this to -Xmx1500M might help. Under 32bit Windows, there is a
maximum of between 1600M and 1700M.
One thing to check for is how much memory is being used by doing
Tools -> Check System Settings
Another thing is to use a profiling tool, see
http://ptolemy.eecs.berkeley.edu/ptolemyII/ptIIlatest/ptII/doc/coding/performance.htm
Someday, I'll take a look at http://bugzilla.ecoinformatics.org/show_bug.cgi?id=4359
Edward wrote:
Actually, the most common source of slowness I've seen is when you
plot very large data sets. The plotter slows down dramatically when
you give it a lot of data...
Original Bugzilla ID was 4360
Also available in: Atom
PDF