IDEA (maybe Java in general) on OSX

I have a general question for the OSX users here that also have some
Windows experience.

It seems to me that IDEA on my G4 with 1.33 GHz and OSX is quite a bit
slower that it was on my old Pentium 3 with 800 MHz and Windows XP.

I wonder if someone else can either confirm that IDEA or maybe Java
(Swing) applications in general are slower on OSX or give some hint on
what to try to make it run faster (maybe some VM settings or something
like that).

It feels slower in overall performance (opening windows, scrolling
through sources, switching editors, etc.)

I am using the latest JDK 1.4.2 that just came out on OSX.

Michael

21 comments

I've been thinking that the latest builds have been feeling faster on my G4 PB - fast enough to make development as pleasant under OS X as it was under XP. I did start experiencing strange general slow-down on my machine a few weeks back that I was only able to resolve by reformatting and reinstalling.

0

Sean Chou <sychou@yahoo.com> wrote:

I've been thinking that the latest builds have been feeling faster on my
G4 PB - fast enough to make development as pleasant under OS X as it was
under XP.


Well, I still feel that it is way too slow. Typing lags behind in medium
sized classes (300 to 500 lines) in my projects. I am not really happy
with that because I consider inputting code one of the core tasks of an
IDE. I hope speed is going to improve even more for the final release.

Dirk Dittert

0

Hello Dirk,
You may give other L&F a try. I find the source scrolling is a bit faster on IDEA default L&F. The overall UI responsiveness of Aqua L&F is a bit slower (I feel pretty satisfied with it on my G4 Cube 400Mhz, though).

0

I have a Powerbook 1.25 Ghz and don't have any complaints with JDK1.4.2. I use the Aqua L&F and life is good. Its still slower than the performance I can get off my Windows boxen.

0

Gregory Pierce wrote:

I have a Powerbook 1.25 Ghz and don't have any complaints with
JDK1.4.2. I use the Aqua L&F and life is good. Its still slower
than the performance I can get off my Windows boxen.


This may be due to the performance of the Java VM on Mac OS X. See
some benchmark numbers on this page: http://www.orbeon.com/oxf/oxfmark

Alex

0

That's pathetic! I hope 1.4.2 fares better. I guess I haven't noticed the difference as much simply because I haven't used IDEA as extensively or daily as I did the Windows version. My experience to date has been decent although it could clearly be a lot better given these numbers. I've not noticed any lag in typing even at large classes. I do notice lags in other areas though.

0

I just ran that benchmark on my G5 (2x2.0 w/ 1 GB RAM) and scored ~2300 (2303, 2297, 2297, 2294).

java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-117.1)
Java HotSpot(TM) Client VM (build 1.4.2-34, mixed mode)

Maybe not first place, but a big jump in the right direction. :)

0

Zoltan Grose wrote:

I just ran that benchmark on my G5 (2x2.0 w/ 1 GB RAM) and scored ~2300 (2303, 2297, 2297, 2294).

java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-117.1)
Java HotSpot(TM) Client VM (build 1.4.2-34, mixed mode)

Maybe not first place, but a big jump in the right direction. :)


ummm most things will have a big jump on a dual G5... that's like a
4ghz pentium 4 or so... of course it's faster... that's like saying, I
was driving an Audi A4 1.8T, and then decided to just try my driving
style with the the Audi S4... there's a big jump in performance hence
the car is improving.... yeah... duh! :)

R

0

It's not the difference with G4 Macs that's relevant here. The dual 2GHz G5 in the table at the link was between 1700 and 1800 with Apple Java 1.4.1.

He tried it on a similar machine with 1.4.2, and got ~2300. That's a 25%-30% increase for just the software upgrade.

Maarten

0

Maarten Hazewinkel wrote:

It's not the difference with G4 Macs that's relevant here. The dual 2GHz G5 in the table at the link was between 1700 and 1800 with Apple Java 1.4.1.

He tried it on a similar machine with 1.4.2, and got ~2300. That's a 25%-30% increase for just the software upgrade.

Maarten


Ah! I see. Sorry missed that. :)

R

0

On 2/9/04 11:12 PM, in article
5958652.1076357531277.JavaMail.itn@is.intellij.net, "Zoltan Grose"
<no_mail@jetbrains.com> wrote:

I just ran that benchmark on my G5 (2x2.0 w/ 1 GB RAM) and scored ~2300 (2303,
2297, 2297, 2294).

java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-117.1)
Java HotSpot(TM) Client VM (build 1.4.2-34, mixed mode)

Maybe not first place, but a big jump in the right direction. :)


I've only got 1750 on the same machine :(
Does that mean there's something wrong with system performance in general?
Can't it be tuned somehow?

--
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"


0

Maxim Shafirov (JetBrains) <max@intellij.com> wrote:

I've only got 1750 on the same machine :(
Does that mean there's something wrong with system performance in general?
Can't it be tuned somehow?


Do you have any server processes running in the background (AFP or
Samba)? Is Entourage running in the background while the test was
running?

Dirk Dittert

0

On 2/10/04 1:48 PM, in article 1g8xvc7.i3d89et0n370N%dittert@despammed.com,
"Dirk Dittert" <dittert@despammed.com> wrote:

Maxim Shafirov (JetBrains) <max@intellij.com> wrote:

>> I've only got 1750 on the same machine :(
>> Does that mean there's something wrong with system performance in general?
>> Can't it be tuned somehow?


Do you have any server processes running in the background (AFP or
Samba)? Is Entourage running in the background while the test was
running?

Dirk Dittert


Samba - yes, Entourage - yes.

--
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"


0

Maxim Shafirov (JetBrains) <max@intellij.com> wrote:

Samba - yes, Entourage - yes.


Try turning off Samba. You could also quit Entourage when running those
tests. I switched to Apple Mail because it seemed to cope better with my
relatively large number of mails.

Another thing could be your energy saving settings. Although I believe
that it should be running at full speed for a desktop mac.

Dirk Dittert

0

Zoltan Grose wrote:

I just ran that benchmark on my G5 (2x2.0 w/ 1 GB RAM) and scored ~2300 (2303, 2297, 2297, 2294).

java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-117.1)
Java HotSpot(TM) Client VM (build 1.4.2-34, mixed mode)

Maybe not first place, but a big jump in the right direction. :)


The 2300 score on a dual G5 2 GHz seems reasonable. There is quite a
boost in performance from the 1.4.1 VM to 1.4.2 in OS X, at least with
this test. And that is a very good news, indeed! This does not
surprise me as it was already mentioned by other people who had a
chance to run the same benchmark with 1.4.1 and 1.4.2 on OS X.

I guess the numbers on the OXF Mark page should be updated to include
results on 1.4.2 for Macs. That would make more sense when doing a
performance comparison with the Windows platform.

Alex

0

I just tested it with a G4 1.33 1GB RAM Powerbook and got 907 with JDK 1.4.2. I have Samba running in the background, but Entourage closed (although the Microsoft Database Daemon is running...whatever that may be).

That appears to be close to a Mobile PIII 1.13GHz - faster than a G5 1.8 with JDK 1.4.1 but slower than a Mobile P4 1.7GHz.

is it me, or have clock speeds lost all meaning? I'm a professional techie (as I'm sure everyone else here is) and I can't keep track of which CPU is comparable to which anymore.

0

>Another thing could be your energy saving settings.
>Although I believe that it should be running at
>full speed for a desktop mac.

My machine started at "Automatic" (Energy Saver panel...Options tab). I set it to Highest just to make sure I got my monies' worth. :)

0

My Powerbook G4 1,33 GHz scores around 900 with JDK 1.4.2_03 (the site
shows around 750 for 1.4.1 on a G4 1,25 GHz). Sound like some
improvement as well but not as much as on your G5.

Zoltan Grose wrote:

I just ran that benchmark on my G5 (2x2.0 w/ 1 GB RAM) and scored ~2300 (2303, 2297, 2297, 2294).

java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-117.1)
Java HotSpot(TM) Client VM (build 1.4.2-34, mixed mode)

Maybe not first place, but a big jump in the right direction. :)

0

Clock speeds are meaningless now unless you couple that with instructions per clock. On many of the new architectures, the pipelines are so incredibly deep and branch prediction has come so far that its more a matter of keeping the CPUs and GPUs fed moreso than their specific clock speeds.

0


Sorry, MHz lost all meaning for me in 1978 when a 4MHz Z80 was subjectively no faster than a 1Mhz 6502 8-)

0

Please sign in to leave a comment.