Closed Bug 107371 Opened 24 years ago Closed 24 years ago

New browser windows take up to 6 seconds to open on an Athlon 1.2Ghz!

Categories

(SeaMonkey :: Bookmarks & History, defect)

x86
OS/2
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: slaughter, Assigned: paulkchen)

Details

From Bugzilla Helper: User-Agent: Mozilla/5.0 (OS/2; U; Warp 4.5; en-US; rv:0.9.5) Gecko/20011016 BuildID: 2001101616 Machine is an Athlon 1.2Ghz with 256MB of RAM, ASUS A7V-E motherboard, running OS/2 MCP with fixpak #1 and kernel 14.080c. Whenever a new browser window is open, either as a pop-up or with CTRL+N, the CPU goes to 100% for 6 seconds before showing the window. Setting CLOCKSCALE=4 in CONFIG.SYS lowered the wait to 3-4 seconds. As a consequence, any page with the usual dozen of popups will take ages to load. Reproducible: Always Steps to Reproduce: 1. Start Mozilla 2. Press CTRL+N or go to an URL that opens pop-ups. Actual Results: The CPU goes to 100% for 6 seconds before showing the window. Expected Results: Well, show the window a little bit faster, being at 1.2Ghz :) So far I have only seen this bug in my machine. A friend with a PIII 850 Mhz opens new windows instantly. "Open New Tab" works perfectly (no wait).
Opening a new window is very slow for me in FreeBSD (PIII 500 and PII 366) as well. Has been for at least a few weeks. It's definitely much slower than it used to be. I'm currently using the 2001102908 linux build, but like I said, I've been seeing this for weeks.
Have you tried it without the new kernel on at all?
I found out I was using a wrong version of the kernel, the W4 version when I was supposed to be using the UNI version (I'm running MCP). Anyway, I tested it with kernel 14.072 and 14.080_UNI. Both showed the same problem. -- Erico
Is there any code that calculates delays in Mozilla or does it rely on system-generated delays? I don't have an Intel CPU of comparable speed to test, but it works on a PIII 850Mhz. Could it be related to the clock speed (or how AMD reports it)? -- Erico
I don't think it has anything to do with AMD. Like I said, I've been seeing this for a while too (and posted about it in the performance newsgroup) and all of the machines I've seen it on have Intel CPUs.
It took about 3s on my p3 (unknown) w/ 128mb of ram running ns6.2 w32 (win98se). wrt popups, 2 suggestions: * disable dom open on load (add it to prefs.js or something) * use tab browser and set it to load new windows in tabs (it's in the preferences dialog in tab browser) - note this requires a semi recent mozilla build (0.9.4/ns6.2 don't have it). what sort of profiling tools can you access? cprof, jprof, and quantify are among the ones we mention on occasion.
Running an Athlon 800 (Linux 2.4.13-ac6), and 0.9.6 takes a long time for new windows to open - perhaps in the range of 6 seconds. 0.9.5 was quite snappy. Using the Debian unstable .debs
Addendum ... I discovered the ``dpkg -i'' operation to install Debian had not been completed (missing dependancy of mozilla-mailnews). When that was fixed by installing mozilla-mailnews, at which dpkg did the configuration for mozilla. It's still a little sluggish to open new windows (~0.7 seconds) but is now at the speed of 0.9.5.
I believe I may have found a possible cause: the size of the "bookmarks.html" file. I tried Monday's nightly build of 0.9.6, and it worked flawlessly with my bookmarks.html file (~500kb), the windows opened instantaneously as they should. Today I downloaded the 0.9.6 release build, installed it on a new directory. With the default bookmarks.html file created, windows open instantaneously. Placing my 500kb bookmark file will make it crawl 6-7 seconds each time a new window is opened. I tested with an older bookmark file (~97kb), and with it new windows take 2 secs to open. Is there anything wildly different between the nightly build and the release build other than debug info/tools? -- Erico
Interesting. There aren't any significant differenced between the nightly and the milestone (except the code changes that took place between the two builds). I'm really not sure how the same bookmarks file (which hasn't changed between the two tests on nightly and milestone build?) could have that impact. Over to Component: Bookmarks for further investigation.
Assignee: asa → pchen
Component: Browser-General → Bookmarks
QA Contact: doronr → claudius
I'm using the following builds to test: 27/11/01 16:47 14243342 0 mozilla-os2-vacpp-0.9.6.zip 28/11/01 15:22 14677713 0 mozilla-os2.zip and these bookmark files: 29/11/01 0:24 498432 0 bookmarks.html 28/11/01 20:53 95614 0 bookmarks.medium 28/11/01 19:27 5225 0 bookmarks.old with the nightly build, windows open instantaneously with any of these. With the release build, the open time appears to be proportional to the size of the bookmark files. I just reinstalled both builds and created fresh profiles to test this. One run with each bookmark file for each build. The situation appears to be consistent. Does the bookmark file get loaded in memory every time a new instance is started or does it use a "shared" copy? -- Erico
Are you still seeing this problem with 0.9.7?
No I'm not. I just keep forgetting to post a comment here, you may close it. 0.9.7 is working fine. -- Erico
Resolving based on reporter's comments.
Status: UNCONFIRMED → RESOLVED
Closed: 24 years ago
Resolution: --- → WORKSFORME
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.