Closed Bug 830200 Opened 13 years ago Closed 12 years ago

Visiting speedtest.net causes plugin-container.exe to peg at 100% utilization when loading

Categories

(SeaMonkey :: General, defect)

SeaMonkey 2.15 Branch
x86
Windows XP
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: thee.chicago.wolf, Unassigned)

Details

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0 SeaMonkey/2.15 Build ID: 20130105222038 Steps to reproduce: I went to http://www.speetest.net to benchmark my Internet connection. Actual results: Plugin-container.exe pegged the CPU at 100% for a good 8 seconds before the page loaded. After loading, the SWF-based benchmarking tool, while idle, was eating 50-60% CPU whereas it used to be next to nothing. Expected results: Plugin-container.exe usually uses only a few % of CPU while idle or while loading the page. Responsiveness of the SWF-based benchmarking tool on the site used to respond lightning-fast and with fluid animation but now it stutters and the animation of the benchmark tool is very cippy. Not sure if it is because of the newer version of Flash (11.5.502.146) or the browser. Went back as far as SM2.13 but the problem was still present. Haven't yet tested it on Flash 11.4.x.
Wrong URL, you want: http://www.speedtest.net/
Cannot confirm. (Though I did get some outrageous times ;-) ) > Plugin-container.exe pegged the CPU at 100% for a good 8 seconds before the page loaded Minimal CPU usage on load & after. Perhaps, at times 15% during. > After loading, the SWF-based benchmarking tool, while idle, was eating 50-60% CPU Before beginning, even when "idle", the tool is doing something. It is displaying a count of "total tests to date". And while doing that, CPU was highly erratic. plugin-container.exe, ~2%, Flash jumping all around, anywhere from a few % to close to 25%. I'm on an i5 750, so a modest system. As a test, try a in a new, clean Profile. If that doesn't help, then go in & disable all your Plugins, except for Flash, & see if there is any change with that. (Above tests were with NoScript, blocking Flash, except at speedtest.net.) --- Restarting in Safe Mode, everything went much the same, but since there is more Flash (from other domains too) going on, I was getting slightly higher & far more consistent numbers (in particular at "idle"). --- I'm actually seeing slightly higher numbers in FF 18, but that's most likely because of differences in ads being served. In any case, can't say I'm seeing anything unusual.
Thanks for fixing the URL. I'm seeing a little bit better performance out of 11.6.602.146 in terms of smoothness of the animation. I am on much older hardware but this didn't use to consume so much CPU with older Flash versions and Firefox. I can only surmise that the Adobe folks are trying to go too far back in terms of GPUs they want to do acceleration with.
Just wanted to report that on SM 2.17b1 and Flash 11.6.602.175, I'm seeing Seamoney.exe peg CPU at load at 99%, then peg out at 80-85% CPU util and plugin-container.exe consuming the rest at 15% CPU util while the test is running. Using NVidia 307.83 drivers from Feb 26th, 2013. I also tested on IE8, fully patched (also on Flash 11.6.602.175), it pegs the CPU at 99% for a few seconds while loading, but once the test loads and I run it, animation is smooth as silk and CPU util is around 2-5%. By all accounts, I am on pretty ancient hardware: Athlon XP 3200, Geforce 6200, 2GB DDR 400, XP SP3 32 bit. This test never had a problem even when I was on a Geforce 5200. At some point I'll try downgrading Flash to 11.2 or something older to see how it behaves.
I went back as far as 11.2.202.235 and the CPU util is the same as with 11.6.x series. IE8 on 11.2.202.235 used a bit more CPU (3-6%) but was overall just fine once the test was running.
Wanted to report that with SM 17b2 and Flash 11.7.700.141 the CPU util has gone down a bit and that the animation seems to be rather smooth and fluid now.
Sorry, that was meant to be SM 2.17b2.
With 2.18b3, plugin-container.exe still pegs at 99% for around 7 seconds while loading the test. This is using Flash 11.7.700.191. Animation seems to be fairly smooth now.
SM2.19 and Flash 11.8.800.88 still pegs CPU at 100% for about 5 seconds for plugin-container.exe on page load. Once loaded, plugin-container.exe fluctuates between 20-65%. It used to peg at 100% for 7 or more seconds but seems to have come down considerably with SM2.19 and 11.8.
SM 2.20 and Flash 11.8.800.129 seems to load the page more quickly but at idle plugin-container.exe still seems to consume a lot of CPU.
Different computer: Celeron 2.40 GHz, 1 GB RAM (768 available) On initial page load, plugin-container went to ~90% - for a good long while, 18s or so. Then it went erratic, jumping around a bit, but up around 65%. Oddly, perhaps, while I ran the test plugin-contain usage went down, while seamonkey.exe went up, perhaps 20%/50%, & then flip flopped a bit. Going to think it is the nature of the beast. (Attempted to put in Flash 10 - without uninstalling Flash 11, but between about:plugins & Addons Manager, I'm not sure what was going on? Like disabling 11 in Addons Manager, also disabled 10 - though that was not reflected in Addons Manager, only in about:plugins. So while 11 is installed, conceivable that you can't "manually" run 10? In the past, you could stick a 10 into a /plugins/ directory & that's all it took. Don't know if there is a plugin.scan.Flash ?)
I was just testing this out with SM 2.22.1 and flash 12.0.0.17beta on a Pentium M 780-based machine and something interesting was happening. Not sure if it is Seamonkey or Flash 12. I am still seeing the large CPU spike on loading this page but it seems to take less time, 15s in my case. Used to be 20s or so. What I am now seeing is that Seamonkey uses 0% CPU while idle after the page has loaded but plugin-container.exe is fluctuating between 5%-80%. Here's where it gets a bit more interesting. If I have the site loaded in a tab and switch to another tab, plugin-container.exe drops to 0% CPU util and stays there. Is it possible the work in bug 847223 has even the slightest thing to do with why i am seeing that?
Well, I am not sure what happened but this bug appears to be fixed and the only thing that changed recently on my system was a bunch of non-WU Hotfixes I put on. I am going to see if I can figure out which Hotfix "fixed" it or if it was a fluke.
Well, in this case, it doesn't seem to be related to a hotfix. Tried with SM 2.19 and then 2.23 and it works. Perhaps the site author updated the SWF file that runs the test to a newer version or something. WFM on XP SP3 and WIn 7 X64 SP1.
Status: UNCONFIRMED → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.