[Thinine]

Originally Posted by Macanoid

to be sure I ran the test again with SP 7 on my macbook 2GHz: Got 5.36 seconds.
did the same test with the latest webkit nightly and got 6.61 seconds

Hmm, there seems to be some performance regressions in WebKit after the point Omni's code is based on. My quad's time is about 2 seconds greater than before, and I'm fairly certain it's because of the first test. Here are my scores:

TEST 1 time: 3.338 sec.
TEST 2 time: 0.495 sec.
TEST 3 time: 0.42 sec.
TEST 4 time: 0.709 sec.
TEST 5 time: 0.243 sec.
TEST 6 time: 1.515 sec.
TEST 7 time: 0.571 sec.

7.29 seconds

[RevEvs]

I wonder why mine takes so long on Test 4, hmmmm oh well.

[Thinine]

Do you have a lot of fonts or do you have any sort of text field plugins or anything?

[Macanoid]

Originally Posted by RevEvs

I wonder why mine takes so long on Test 4, hmmmm oh well.

With OW 5.5 or with Safari? If the latter, which version and did you try the nightly build webkit?

[RevEvs]

Hmmm as suspected - I had some SIMBL bundles, and even though i had them in a disabled folder it seems they still (even though not loaded) seemed to effect things - i'm down to the 6second mark now. cool cool.

[doublep]

Sneaky Peek v9 is now out!

[Macanoid]

all the sneaky peek run great except for one thing, they gobble up memory like there's no tomorrow. Last I checked OW used 295 mb real memory ?!?!?

[Horsepoo!!!]

Originally Posted by Macanoid

all the sneaky peek run great except for one thing, they gobble up memory like there's no tomorrow. Last I checked OW used 295 mb real memory ?!?!?

I wouldn't be surprised...WebKit is by far the leakiest thing I've ever seen. Even after seemingly total rewrites, the thing leaks like there's no tomorrow in the latest nightly builds.

I don't know which build Omni picked but chances are it was just as leaky as any other WebKit build considering this problem has been plaguing the Safari team for over a year now.

I don't know what the **** the Safari team is doing but they seem to be wasting a lot of time. The code clean up hasn't been very useful as far as plugging leaks and they've been working on some cross-platform and CSS-compatible text input for the past 2 months now. Jeebus.

[Macanoid]

That's really bad news, if true. Last night I also checked Safari's memory use after having it open for about 6 hours and found it used nearly 200 MB of real memory too ?!?!? For f*ck's sake, 200 mb for just a browser. I suppose I might just as well switch back again to OW as it's faster and has a nicer feature set

[RevEvs]

Well I knew there was a reason I got 2.5Gig of RAM all those years ago - to run OmniWeb

[Catfish_Man]

Originally Posted by Horsepoo!!!

I wouldn't be surprised...WebKit is by far the leakiest thing I've ever seen. Even after seemingly total rewrites, the thing leaks like there's no tomorrow in the latest nightly builds.

You know that rewrites usually increase, not decrease, leaks. Right?

<edit>
As well as general bugginess. Leaks and trivial bugs are best fixed via incremental revision.
</edit>

[Thinine]

What rewrite? And what leaks? If you find some leaks, report them. http://webkit.org

[Thinine]

WebKit's performance problem in the latest builds can be fixed by the following in Terminal: defaults write com.apple.Safari WebKitThrottleWindowDisplay YES

There was a conflict between WebKit's repainting and Safari's, which cause the visible slowdown. My Quad's time is back down to 5.19s.

[Catfish_Man]

Also, http://bugzilla.opendarwin.org/show_bug.cgi?id=8844 was just filed on the leaks.

[Thinine]

Those are from a patch in the last day, so not what they're talking about.

[workerbee]

SP 10 is now out.

[cpac]

holy memory leaks batman. Leave it open for 30 mins+ on my PB and the free space (2.4GB) goes down to near nothing and OSX warns me that I'm running out of hard drive space.

Still, it's nice to have a snappy (there, I said it) version of OW again.