I have an
Athelon pro 2000 1300 MHz clock
256 Meg ram
Win98SE
Boinc 4.09

I have been running Seti and LHC for a while now with no problems.
I re-attached to Climate Prediction this morning.
When I left the computer at 14:00 all seemed to be running soothly with LHC doing it thing.
I returned around 16:40 and noticed that both Seti and Climate Prediction appeared to be running simultaneously.
At 16:51 the process changed from Seti to LHC as expected , but Climate Prediction seem to continue to tick away.
I allowed this to continue for half an hour noticing that LHC was progressing at about a tenth of its normal speed.
I then shut down Boinc and restarted it. Only LHC started, but it was obvious that it was still running much slower than it should.
I rebooted the computer and LHC has run at its expected speed and now a switch to Seti has occurred normally.

I am not very computer literate, but I hope someone can make sense of this.


This is part of this afternoons log to the point where I first closed down the program..

LHC@home - 2004-10-01 12:01:40 - Started upload of v64lhc1000pro39s6_8515.39_1_sixvf_122064_1_0
LHC@home - 2004-10-01 12:01:55 - Finished upload of v64lhc1000pro39s6_8515.39_1_sixvf_122064_1_0
LHC@home - 2004-10-01 12:01:55 - Throughput 3134 bytes/sec
LHC@home - 2004-10-01 12:18:43 - Sending request to scheduler: http://lhcathome-sched1.cern.ch/scheduler/cgi
LHC@home - 2004-10-01 12:18:47 - Scheduler RPC to http://lhcathome-sched1.cern.ch/scheduler/cgi succeeded
LHC@home - 2004-10-01 12:18:47 - Project prefs: no separate prefs for home; using your defaults
--- - 2004-10-01 12:18:56 - Suspending network activity - user request
SETI@home - 2004-10-01 12:57:11 - Pausing result 26ap04aa.4387.9858.154822.93_3 (removed from memory)
climateprediction.net - 2004-10-01 12:57:12 - Starting result 31gr_000163297_3 using hadsm3 version 4.04
climateprediction.net - 2004-10-01 13:57:12 - Pausing result 31gr_000163297_3 (removed from memory)
LHC@home - 2004-10-01 13:57:14 - Starting result v64lhc1000pro46s6_8564.89_1_sixvf_129607_0 using sixtrack version 4.46
LHC@home - 2004-10-01 14:51:43 - Computation for result v64lhc1000pro46s6_8564.89_1_sixvf_129607 finished
climateprediction.net - 2004-10-01 14:51:44 - Restarting result 31gr_000163297_3 using hadsm3 version 4.04
SETI@home - 2004-10-01 15:51:44 - Restarting result 26ap04aa.4387.9858.154822.93_3 using setiathome version 4.05
climateprediction.net - 2004-10-01 15:51:44 - Pausing result 31gr_000163297_3 (removed from memory)
SETI@home - 2004-10-01 16:51:44 - Pausing result 26ap04aa.4387.9858.154822.93_3 (removed from memory)
LHC@home - 2004-10-01 16:51:44 - Starting result v64lhc1000pro46s6_8562.55_1_sixvf_129581_1 using sixtrack version 4.46

If you have Keep Processes in Memory when Paused set to yes, on 95/98/ME machines the processes that have been paused can still get some CPU time as the code for halting a process temporarily is not as robust on these operating systems as it is on NT/2K/XP/2003.

I saw the same thing happening on my linux box. Seti and LHC were both getting 49.9% of the CPU. But when BOINC tried to re-activate seti, seti crashed and had to restart computation on the work unit. I upgraded to BOINC 4.11 (beta) and it seems to have fixed it. There were some fixes to the CPU scheduler going from 4.09 to 4.11. Note that 4.11 is not an official release yet so run at your own risk if you want to try it.


---------------------------------------
- A member of The Knights Who Say NI!
Possibly the best stats site in the universe:
http://boinc-kwsn.no-ip.info

> If you have Keep Processes in Memory when Paused set to yes, on 95/98/ME
> machines the processes that have been paused can still get some CPU time as
> the code for halting a process temporarily is not as robust on these operating
> systems as it is on NT/2K/XP/2003.

As indicated in the log Keep Processes in Memory when Paused is set to NO and normally should have released memory.

> I saw the same thing happening on my linux box. Seti and LHC were both
> getting 49.9% of the CPU. But when BOINC tried to re-activate seti, seti
> crashed and had to restart computation on the work unit. I upgraded to BOINC
> 4.11 (beta) and it seems to have fixed it. There were some fixes to the CPU
> scheduler going from 4.09 to 4.11. Note that 4.11 is not an official release
> yet so run at your own risk if you want to try it.

I'm waiting to see if it happens again. Climate prediction just finished its cycle has I was answering your message. It seem too have cleaned up after itself, as the Seti process in running on its own at what appears to be the right speed.

Perhaps I'll wait a little longer before taking a chance on 4.11 (Although I did get lucky with 4.09 Alpha).

> I saw the same thing happening on my linux box. Seti and LHC were both
> getting 49.9% of the CPU. But when BOINC tried to re-activate seti, seti
> crashed and had to restart computation on the work unit. I upgraded to BOINC
> 4.11 (beta) and it seems to have fixed it. There were some fixes to the CPU
> scheduler going from 4.09 to 4.11. Note that 4.11 is not an official release
> yet so run at your own risk if you want to try it.
>

>

I too have seen it on linux...
If the fix is there in 4.11, it would be kind of nice if the devs mentioned this was even an acknowledged problem, even if the fix isn't available to the masses.

There's been a thread under the "q+a" section about this for a number of days (linux).