Message boards :
Number crunching :
4.3x issue -- fighting to regain control from screensaver
Message board moderation
Author | Message |
---|---|
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
I've noticed this problem since installing 4.36. It was there to some degree in 4.19 also, but is now more of a problem. I'm currently running 4.38, and it is still happening. When the screensaver comes on, sometimes I have to fight it to regain control of the computer. This happens especially if the machine is doing something CPU intensive, like encoding OGG files, say. When I move the mouse, sometimes nothing happens for a while, then I get a glimpse of the desktop, followed by a return to the screensaver, then after a while, another glimpse of the desktop, followed by another return to the screensaver, etc. etc. Eventually, the screensaver gives up and I have control again after many cycles of this. Maybe it's my imagination, but it appears to go on longer if the screensaver has been running for a longer period of time. |
keputnam Send message Joined: 2 Jul 99 Posts: 242 Credit: 2,736,564 RAC: 3 |
While not exactly the same situation, I've noticed with the v4.3x CCs, that I usually have to Pause Boinc to get another app (Word, Firefox, Task manager- basically any other app) to initialize. 4.32 would resume fine, but 4.35, 4.36 and 4.37 all say they have resumed and one of Science Apps is listed as running, but it does not have any time in task Manager. I have to stop the BOINC service and restart it to get Boinc processing again |
stomp Send message Joined: 4 Sep 99 Posts: 8 Credit: 238,321 RAC: 0 |
I have some what of the same problem, but what I notice, is that LHC project (sixtrack 4.67) is running each time (this happens on two of my machines (HT and a mobile)). But this only happens when LHC is the active project. On the HT machine both threads would be running LHC. |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
Update -- it seems to happen only on slower (read less than 2.0GHz) machines, dual or not. I've had it happen on every one of my less than 2.0 GHz machines. (there was more to this post, but I had a "less than" sign above, and everything after it was lost) :( ------edit----- Interesting note -- may or may not be related -- when I first installed 4.38 on my 3.20GHz HT machine, it started up two CPDN WU's. When the screensaver kicked in, there was a lot of flashing between the graphics and black at about 5-10 times a second. Moving the mouse stopped it. Same thing happened next time. I stopped the Manager and the CC, then restarted, and this behavior did not reappear. Freak occurrence? |
stomp Send message Joined: 4 Sep 99 Posts: 8 Credit: 238,321 RAC: 0 |
My HT is a 3.0GHz HT, so its not just slow machines. |
Rom Walton (BOINC) Send message Joined: 28 Apr 00 Posts: 579 Credit: 130,733 RAC: 0 |
LHC isn't notifing the daemon that it shutdown due to user input, so the daemon automatically starts up a new application believing the old one just completed it's workunit. We are working on the issue. ----- Rom BOINC Development Team, U.C. Berkeley My Blog |
stomp Send message Joined: 4 Sep 99 Posts: 8 Credit: 238,321 RAC: 0 |
Thanks for the update!!! |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
I could have sworn it was more than just LHC, but LHC has done this for a long time. Thanks, Rom, for the update. I will report back if I can verify another project doing this. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
<blockquote>I could have sworn it was more than just LHC</blockquote> This is listed as the "Known Issues" related to 4.38. Taken from the Download page. Known issues If BOINC runs at the same time as Windows XP 3D screensavers, the system becomes sluggish and unresponsive. Same issues for dialup users as 4.25. Same issues for ATI graphics cards as 4.25. Does this apply to the problem? |
Rom Walton (BOINC) Send message Joined: 28 Apr 00 Posts: 579 Credit: 130,733 RAC: 0 |
Not really, in this case BOINC is the selected screensaver, not one of the Windows XP 3D screensavers. ----- Rom BOINC Development Team, U.C. Berkeley My Blog |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
It just happened again, this time with 4.39 that I just installed. The machine was a Dual P3, 933MHz. There were only 3 unfinished WU's left on the machine, one CPDN and two Einstein. The CPDN WU was running and one of the Einies. The Einstein graphics were running, and the client decided to switch to the CPDN graphics. In somewhat less than a minute, I moved the mouse. The CPDN graphics went away, the desktop came back, and a second or two later, the Einstein graphics came back. I moved the mouse again, and the desktop came back for good. An easy one. If the screensaver had been running for a while, I'd bet it would have been harder (from past experience). Here's the kicker: I looked at the messages, and NOTHING was in the log at that time. i.e. no switchover from one wu to another. Since CPDN was already running, the only possibility was to stop one Einie and start the other. Neither was anywhere near finished. What apparently happened, was that BOTH graphics were started, and they were competing with each other. Another interesting note -- none of my machines with 4.38 were doing it recently, they seem to have settled down. It just happened after the new install of 4.39 about an hour ago. |
Rom Walton (BOINC) Send message Joined: 28 Apr 00 Posts: 579 Credit: 130,733 RAC: 0 |
4.40 should contain the fixes for this issue. David and I talked over the solution to the screensaver issue after I had already released 4.39, given the nature of the bug I thought it best to get it out there sooner rather than later. ----- Rom BOINC Development Team, U.C. Berkeley My Blog |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
Thanks, Rom. I haven't seen it since upgrading to 4.40 on 6 machines so far. Great work! |
stomp Send message Joined: 4 Sep 99 Posts: 8 Credit: 238,321 RAC: 0 |
The question asked if it applied to the "Know Issues", No it does not apply. But I did load 4.40 and will monitor that on the two machines that was having problems. Thanks Reference to another thread: Warning! Only use Development Version if you like to help test things out! |
Ace Casino Send message Joined: 5 Feb 03 Posts: 285 Credit: 29,750,804 RAC: 15 |
Everyone wants faster crunch times. Your cpu is running your screen saver. I turned my screen saver off a couple years ago with Seti Classic and found my time to complete a WU dropped big time. Even when your monitor turns off your screen saver is running using cpu power. I'm no expert but if you have a screen saver running either Seti or something else it's using processor power. I suspect someone will tell me it's only using a very small amount of processer power but it sure made a big difference on my computer. If you want to see the graphics all you have to do is click on the WU that is crunching and click on "show graphics". I just have my monitor turn off after 5 minutes...NO screen saver. |
genes Send message Joined: 25 May 99 Posts: 117 Credit: 580,187 RAC: 0 |
> Even when your monitor turns off your screen saver is running using cpu > power. I'm no expert but if you have a screen saver running either Seti > or something else it's using processor power. If you set the screensaver to "go to blank screen", an option with BOINC, then it stops using CPU power after the time you set. This is independent of the option to turn off monitor power (Windows option) found under the screensaver tab. The BOINC screensavers are written using OpenGL and use mostly the display card's GPU, not your CPU. The original SETI screensaver was not, though it did have the option to go blank. ----- update -- still have not had the "fighting with screensaver" issue return since 4.40, but 4.40 has other issues, so beware. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.